Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Effect of the geometric form and meaning of the stimulus on the configuration of the visual evoked response Purves, Sherrill Jean Swift 1976

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1976_A1 P87.pdf [ 5.75MB ]
Metadata
JSON: 831-1.0093878.json
JSON-LD: 831-1.0093878-ld.json
RDF/XML (Pretty): 831-1.0093878-rdf.xml
RDF/JSON: 831-1.0093878-rdf.json
Turtle: 831-1.0093878-turtle.txt
N-Triples: 831-1.0093878-rdf-ntriples.txt
Original Record: 831-1.0093878-source.json
Full Text
831-1.0093878-fulltext.txt
Citation
831-1.0093878.ris

Full Text

THE EFFECT OF THE GEOMETRIC FORM AND MEANING OF THE STIMULUS ON THE CONFIGURATION OF THE VISUAL EVOKED RESPONSE by SHERRILL JEAN SWIFT PURVES B.Sc, McGill University, 1967 M.D. University of B r i t i s h Columbia, 1971 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY i n THE DEPARTMENT OF PHYSIOLOGY We accept this thesis as cxsnforining to the required standard THE UNIVERSITY OF BRITISH COLUMBIA May, 1976 In p r e s e n t i n g t h i s t h e s i s in p a r t i a l f u l f i l m e n t o f the requirements f o r an advanced degree at the U n i v e r s i t y of B r i t i s h Columbia, I agree that the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e fo r reference and study. I f u r t h e r agree t h a t permiss ion fo r e x t e n s i v e copying o f t h i s t h e s i s f o r s c h o l a r l y purposes may be granted by the Head of my Department or by h i s r e p r e s e n t a t i v e s . It i s understood that copying or p u b l i c a t i o n of t h i s t h e s i s f o r f i n a n c i a l ga in s h a l l not be a l lowed without my w r i t t e n p e r m i s s i o n . Depa rtment The U n i v e r s i t y of B r i t i s h Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1W5 ABSTRACT In the visual system i t has been established that certain physical parameters of the stimulus including intensity, focus, and size of the elements of a pattern have significant effects on the configuration of the evoked response as recorded from the o c c i p i t a l region. While claims have been made in a few publications, to this date the question of whether any part of the evoked response i s affected by higher perceptual processes rather than ones more directly related to stimulus input has not been resolved. Evoked responses to four geometric shapes (a square, c i r c l e , e l and omega) were recorded from multiple scalp locations under two experimental conditions; i n the f i r s t the shapes were presented at random intervals and i n random order with no meaning assigned to them. In the second they were presented i n a fixed rhythmic sequence, and two of the shapes occasionally f a i l e d to appear i n their usual time interval i n the sequence. The subject"was required to rapidly signal such omissions by a button press. The cerebral e l e c t r i c a l a c t i v i t y during this time interval of the expected, but omitted stimulus was recorded and separately averaged, and these responses were called emitted potentials. There were significant, differences between the evoked responses (in the oc c i p i t a l region) to the square and e l shapes and between those to the c i r c l e and omega shapes. These differences were demonstrated by three measurement techniques: performance of the discriminant functions computed by SWDA i n classifying single t r i a l responses, a ratio s t a t i s t i c called X as described by John, Herrington & Sutton (1967), and amplitude differences i n the N 2 and P 2 components. The emitted potentials recorded from the vertex, contained a small and variable early negative component and an obvious late positive component similar to the one seen i n the responses to the present square and e l during the second experimental condition. Manipulation of the waveforms from the single t r i a l s prior to averaging showed that this negative component was not time-locked to the peak. The evoked potential differences found were believed to be due to two classes of variables; the physical characteristics of the stimulus (the contours i n the central 1.5° of the visual field) and task-related changes i n the meaning of the stimulus; and these affected e a r l i e r and later parts of the waveform respectively. TABLE OF CONTENTS: Page t Abstract i i Table of Contents . i v L i s t of Figures v i i L i s t of Tables v i i i L i s t of Abbreviations i x Acknowledgement x I. Introduction 1 II. Review of the Literature 5 2.1 The relationship of scalp-recorded evoked responses to the underlying brain a c t i v i t y 5 2.2 Pattern VER's i n the study of vision 7 2.3 Evoked responses i n the study of psychological variables and meaning 14 2.4 Suitniary of the literature review 20 III. Statement of the problem for investigation 22 IV. Methods 24 4.1 Data acquisition methods 24 4.2 Data analysis methods 33 V. Results 38 5.1 Effect of different experimental conditions on the evoked response 38 5.2 Effect of different stimulus shapes on the evoked responses 42 5.2.1 Results of Visual analysis 43 5.2.2 Results of Stepwise discriminant analysis 50 5.2.3 Results using the X descriptor computations 60 5.2.4 Summary of the effect of the different stimulus shapes on the evoked responses 64 5.3 Emitted responses 67 5.3.1 Averaged emitted responses 67 5.3.2 Averaged "shifted" emitted responses 72 5.3.3 Control studies: motor potentials, EOG contributions, and topography . . . 76 5.3.4 Summary of findings for omitted stimuli 79 VI. Discussion 81 6.1 Studies of evoked responses and stimulus meaning 81 6.2 Techniques for the analysis of evoked responses 83 6.2.1 Visual analysis 85 6.2.2 X values 85 6.2.3 Stepwise discriminant analysis 86 6.3 Topographical distribution of the evoked responses 89 6.3.1 Location of shape effects 89 6.3.2 Distribution of response components.. 90 v i 6.4 Intershape differences related to physical properties of the stimuli 93 6.5 Effects of stimulus meaning 99 6.5.1 Present stimuli 99 6.5.2 Omitted stimuli 101 VII Summary & Conclusions 105 Bibliography 110 Appendix A. Computer hardware and data acquisition program (VER03) 118 Appendix B. Stepwise Discriminant Analysis 120 Appendix C. Terms from experimental psychology .. 124 vii. LIST OF FIGURES Figure Page 1. Electrode placements 25 2. Data acquisition system 27 3. Stimulus shapes 29 4. Experimental paradigms 30 5. Effect of experimental conditions for circle and omega . 39 6. Effect of experimental conditions for square and e l .... 41 7. Effect of stimulus shape on the evoked responses 44 8. Intersubject variability; occipital VER to e l for 13 subjects 45 9. Effect of stimulus size 52 10. Distribution of latency points chosen by the 4-group discriminant functions 57 11. Distribution of latency points chosen by the 2-group discriminant functions 62 12. Distribution of X values 65 13. Averaged emitted potentials at C in five subjects 68 14. Averaged emitted potentials at multiple recording sites in one subject 69 15. "Shifted" averaged emitted potentials at C z in five subjects 73 16. "Shifted" averaged emitted potentials at multiple recording sites in two subjects 74 17. Responses to omitted stimuli with EOG control 77 18. Responses to omitted stimuli in frontal regions with absent button press 78 19. Flow chart of VER03 program (Appendix A) 119 v i i i LIST OF TABLES: TABLE I. Latencies of the components of o c c i p i t a l evoked responses to the four shapes 48 I I . Amplitudes of the components of o c c i p i t a l evoked responses to the four shapes 50 III. Performance of the "four-shape" discrinunant functions 54 IV. Percentages of t r i a l s incorrectly c l a s s i f i e d separated into pairs of shapes confused with each other 56 V. Performance of the "two-shape" discriminant functions. One subject, five pairs of shapes 59 VI. Performance of the "two-shape" discrinunant functions. Five subjects, two pairs of shapes 61 VII. Mean A values for each subject 63 VIII. Amplitude and latency of P^ for responses to omitted and present stimuli 70 IX. Amplitude and latency of N 2 for responses to omitted and present stimuli 71 LIST OF ABBREVIATIONS EEG Electroencephalogram EOG Electro-oculogram EP Evoked potential ER Evoked response SWDA Stepwise Discriminant Analys VER Visual Evoked Response X The author gratefully acknowledges the assistance in this research work provided by a number of people, especially Dr. Mort Low whose advice, encouragement, and t i r e l e s s personal involvement i n the project were invaluable for me Michael Baker for his interest and a b i l i t y i n the design and writing of the computer programs for the PDP-11 system Janice Galloway and Richard Ferguson for their technical assistance with the experimental apparatus and the subjects Jim McEwen and John Doyle from the Dept. of E l e c t r i c a l Engineering for their helpful instruction i n the use of the UBC Computing Centre F a c i l i t i e s Dr. Michael Schulzer from the Dept. of Mathematics for advice on s t a t i s t i c a l testing and my committee members i n the Dept. of Physiology; Drs. Tony Pearson, Hugh McLennan and Peter Vaughan for their consideration, suggestions and criticisms of the project. Financial support was provided by the Medical Research Council of Canada through an MRC Fellowship to the author from September 1972 u n t i l December 1975 and MRC Grant MT-3313. Chapter I INTRODUCTION To attempt to understand the enormous complexity of neural transactions within the human brain i s a formidable challenge to investigators i n a l l of the disciplines of neurological research. One of the methods of directly studying the neural processes i n the brain, i s to record the e l e c t r i c a l a c t i v i t y occurring i n response to sensory stimulation. Two classes of e l e c t r i c a l a c t i v i t y can be recorded; the minutely localized potentials of single nerve c e l l s and the more diffuse and slowly changing potential f i e l d s detectable i n the v i c i n i t y of large populations of neurons. The terms "evoked potentials" and "evoked responses" are coimonly used to denote the signals of the lat t e r class of cerebral e l e c t r i c a l a c t i v i t y that are recorded following stimulation of sensory receptors or afferent pathways. A stimulus i n i t i a t e s a sequence of physiological events, which are the substrates for i t s perception, as well as processes leading to an overt behavioral response. Therefore, analysis of the e l e c t r i c a l a c t i v i t y occurring between stimulus and response should provide clues to the nature and anatomical location of these physiological events. Since the late 1950's, there has been a large amount of data published about the effect of a wide variety of stimulus parameters on the evoked responses recorded from human subjects. The stimulus parameters investigated have included both physical characteristics such as intensity, 1 2. duration and size, and psychological-contextual ones such as expectation, task-relevance and stimulus meaning. With respect to the la t t e r , conflicting reports have appeared i n the literature. In spite of their importance for the development of theories of information processing, some widely publicized findings on the effect of stimulus meaning on the evoked potential configuration have been poorly substantiated. The use of patterned visual stimuli to e l i c i t evoked responses i s a recent development i n this f i e l d of study, because of the relative technical complexity of the stimulus presentation apparatus (as compared to the simple stroboscopic flash that was used previously to stimulate the visual system). With patterned stimuli the experimenter can provoke and examine the physiological events involving more complex aspects of visual perception and related cognitive processes than was formerly possible using only the flash. Another recent advance i n the study of cognitive processes related to processing of sensory information, appears i n some recent reports of the recording of a consistent wave form from the scalp i n the interval when a subject expects but i s not actually presented with a stimulus. The responses recorded i n the interval when no stimulus has been presented are called emitted potentials, to distinguish them from the evoked potentials e l i c i t e d by stimuli that are actually presented. The study of these emitted potential wave forms may provide important information concerning brain processes without interference from the acti v i t y d i r e c t l y related to the stimulus input. 3. In recordings made from the scalp, the evoked response i s of much smaller amplitude (l-20yv) than the ongoing electroencephalographic (EEG) acti v i t y which i s 10 - 100 yv i n amplitude. In order to isolate and study the act i v i t y related to the stimulus, repeated responses are averaged starting at the time of stimulus presentation. The event-related ac t i v i t y which, i s presumed to be relatively constant i n the repeated t r i a l s i s enhanced and the background non-event related EEG i s cancelled by this method. The number of t r i a l s per average has varied widely i n different reports, ranging from 10 to 2000 or more repetitions of the stimulus. The variance of these t r i a l s included i n the average i s rarely ever recorded. The resulting averaged evoked responses can be examined. The peaks are labelled and their latency and amplitude can be measured. When simple stimulus variables such as luminance are altered, the effect i s usually only seen i n one of the peaks or components of the evoked response and i s f a i r l y easy to observe and measure. However when changes are made i n complex stimulus variables such as meaning, more than one component i n the waveform may be altered and i t becomes quite d i f f i c u l t to quantify differences i n the waveforms that can be related to stimulus variables. If the evoked EEG a c t i v i t y i s digitized and then treated as a series of discreet observations, some quite different methods, other than averaging can be used to approach this problem. One such method i s called discriminant analysis; i t i s a technique based on multivariate s t a t i s t i c a l theories that determines the variables (or time points) that best account for the differences between two or more sets of observations (waveforms). 4. It requires very complex computations and must be carried out on a large computer. (A program called BMD:07M Stepwise Discriminant Analysis from the UCLA series i s available i n most computing program l i b r a r i e s ) . In order to investigate the nature and anatomical location of the (cortical) events that underlie visual perception, a study was designed using complex patterned visual stimuli (geometric shapes) to demonstrate what, i f any, differences i n the evoked potential configuration could be attributed to differences i n stimulus meaning or to differences i n the physical properties of the stimuli. The experimental design also included a paradigm to record the emitted potential a c t i v i t y that appeared during the interval when the subject expected but did not see one of the geometric shapes. A l l of the recorded waveforms were digitized and stored as single t r i a l s so that a variety of techniques, including averaging and discriminant analysis could be used to measure any differences between them. 5. Chapter II REVIEW OF THE LITERATURE 2.1 The relationship of scalp recorded evoked responses to the underlying brain a c t i v i t y . A f a i r l y comprehensive review of the experimental evidence and current theories i s provided by Kiloh, McComas & Osselton (1972) for the origin of spontaneous a c t i v i t y and by Regan (1972) for the origin of both spontaneous and evoked EEG a c t i v i t y . The available data about the origins of scalp-recorded e l e c t r i c a l potentials are fragmentary and do not yet precisely define the sources of either spontaneous EEG a c t i v i t y or evoked potentials. Early reports suggested that the 0-60 Hz. "slow" a c t i v i t y i n the EEG recorded at the c o r t i c a l surface i s the result of a summation of action potentials of c o r t i c a l c e l l s . Others later showed conclusively that neuronal action potentials were not essential and suggested that the waves of the surface EEG result from a summation of the excitatory or inhibitory postsynaptic potentials developed by the soma and larger dendrites of the c o r t i c a l pyramidal c e l l s (Li & Jasper, 1953). This suggestion received further support from studies including intracellular recordings of c o r t i c a l neurons (Jasper & Stefanis, 1965; Creutzfeldt, Watanabe & Lux, 1966) that showed a close time relationship between repetitive postsynaptic potentials and macro potentials recorded on the surface of the overlying cortex. In some more recent work E l u l (1968) demonstrated the existence of slow, spontaneous membrane potential shifts i n c o r t i c a l neurons, i n the same frequency range as the EEG. On the basis of these observations he suggested 6. that the EEG may r e f l e c t these slower potentials rather than being a temporal summation of faster a c t i v i t y , as the e a r l i e r studies of post-synaptic dendritic potentials had implied. The relationship of the EEG and evoked potential a c t i v i t y recorded from the scalp to the e l e c t r i c a l a c t i v i t y recorded from the c o r t i c a l surface has been investigated by a number of workers. Geisler & Gerstein (1961) showed both experimentally (in monkey) and i n theoretical formulations that a major attenuation of e l e c t r i c a l potential amplitude occurs i n the highly conductive layers of the CSF and dura but not at the skull or scalp. Cooper, Winter, Crow & Walter (1965) i n studies on human patients with simultaneous recording from the scalp and implanted c o r t i c a l electrodes, found that only "widely synchronized" c o r t i c a l a c t i v i t y was not obscured at the scalp and that the amount of attenuation between cortex and scalp depended upon the size of the area of cortex involved i n the ac t i v i t y . In a study of the EEG a c t i v i t y following sensory (flash) stimulation, Heath & Galbraith (1966), using c o r t i c a l and scalp electrodes i n the primary visual area showed that an early component (65 msec.) was evident at the same latency i n both c o r t i c a l and scalp recx)rdings; but for later (100 to 300 msec.) peaks there were latency differences between the response components recorded at the two electrodes. In explanation of these waveform differences, they suggested that a c t i v i t y generated i n non-primary cortex spread to the o c c i p i t a l scalp lead but not to the c o r t i c a l lead. Hence they concluded for VER's to flash, that the early components in the response recorded from the scalp electrode dir e c t l y over the primary visual cortex are generated only in the immediately underlying brain area, while the 7. later response components represent summation of a c t i v i t y from more extensive (including both primary and non-primary cortex as far away as the temporal area) brain regions. Vaughan (1966) reported similar findings i n a patient explored with c o r t i c a l electrodes during an intracranial surgical procedure. However, i n studying the distribution of evoked response to flash i n the c o r t i c a l regions adjacent to the calcarine fissure, he found the response differed markedly as the recording electrode was moved anteriorly from the o c c i p i t a l pole. He concluded that most of the scalp VER usually recorded i n the o c c i p i t a l regions originated i n the o c c i p i t a l pole of the cortex (foveal representation) and not i n more anterior portions of the striate cortex. This finding i s consistent with other authors' observations (Devoe, Ripps & vaughan, 1968; Harter, 1970) that the VER primarily reflects foveal stimulation. 2.2 Pattern VER's i n the study of vision. The response recorded i n the o c c i p i t a l region e l i c i t e d by spatially structured stimulus fi e l d s i s quite different from the one evoked by a flash. This major difference was f i r s t described by Sphelmann (1965) who showed that a patterned stimulus evokes a response of much greater amplitude and containing peaks of different latency and polarity than a flash of comparative l i g h t intensity. The type of pattern presentation used to e l i c i t the "pattern visual evoked response" i s important. The stimulus may be a brief presentation of a pattern on a blank background; i t may be a change of pattern from one form to another (called a "pattern reversal" i n the literature) or i t may 8. be illumination of a dark pattern by a flash. The theoretical advantage of a pattern reversal type of stimulation i s that there i s no change i n the t o t a l luminance reaching the subject's eye, only a change i n distribution of pattern structure within the visual f i e l d . Consequently some workers (Halliday & Michael, 1970) have suggested that the response to pattern reversal contains only "pattern specific" components, and not any to luminance change as does the flash VER. However, the pattern reversal method i s usable only for certain types of symmetrical patterns such as checkerboards or gratings. I t has also been shown by Estevez & Sprekreijse (1974) that the potential generated by pattern reversal includes components found both i n responses to a pattern appearance and a pattern disappearance (analgous to an "on" and "off" response), and hence i s not as uncomplicated as ori g i n a l l y suggested by Halliday. The third method of presenting the patterned stimulus (by illurninating a dark pattern with a stroboscopic flash) requires less elaborate instrument-ation and has been used i n many studies including the original one of Sphelmann (1965). The VER produced by this type of stimulation consists of a summation of the responses to both the large luminance change produced by the flash and to the patterned stimulus. For a l l of the above stimulation techniques the stimuli are sufficiently widely spaced i n time that the nervous system has been regarded as returning to the same i n i t i a l state between successive stimuli, and hence these 9. responses are sometimes called "transient" responses to distinguish them from another type of evoked potential described i n the literature called a "steady state" evoked response. (See Regan (1975) for a complete description of this phenomenon). The term "steady state" i s employed to indicate that the stimuli are presented at such a rapid rate (e.g. 10 - 60 Hz.) that the co r t i c a l responses to individual stimuli overlap. The steady state response i s then separated from the non-event related background EEG because i t i s composed of components (or harmonics) whose frequencies are exactly the same as or multiples of the stimulus frequency. A Fourier analysis i s used to perform the required computations and the amplitude and phase of the frequency components can be displayed and measured. The technique has the advantage of speed because the stimuli can be presented so much more rapidly than they can be for the study of transient evoked responses. Regan (1974) has suggested that the properties of the components of the steady state response seem to depend on whether the stimulation frequency i s (a) near 10 Hz., (b) 13 - 25 Hz., or (c) 45 - 60 Hz. and has found that some diseases involving the visual system may affect these frequency bands differently. The steady state technique has also been used by Campbell & Kulikowski (1972) i n studies of the relationship of a subject's a b i l i t y to see a sine wave grating pattern of low contrast to the size of the evoked response e l i c i t e d by the same grating pattern. They separately tested the subject's threshold by having him report when the grating pattern disappeared as the contrast was decreased, and then by recording the evoked-response to a rapid pattern reversal of the grating at different levels of contrast. They found good agreement between the threshold determined by the 10. subject's verbal report and the contrast level at which the evoked response could no longer be detected. The steady state pattern evoked response thus provides an objective indicator of visual function and perception. Transient and steady state evoked responses give complementary insights into brain function. Van Hof (1960) and Tweel & Verduynlunel (1965) showed that i t i s not possible to predict the parameters of the steady state responses from the characteristics of the transient responses. This i s not surprising i n view of the very different stimulation techniques, but consequently results obtained with one method are not directly comparable with those obtained with the other. The remainder of the literature on pattern VER's to be reviewed i s concerned with the transient type of response, as i s the experimental work described i n this thesis. The individual peaks of the transient pattern VER have been studied using topographical mapping techniques i n an attempt to determine the area of cortex i n which these components originate. Jeffreys (1971) and Jeffreys & Axford (1972a & b) found that the f i r s t two peaks of the VER occurring 65 - 80 msec, and 90 - 110 msec, after presentation of a checkerboard pattern (they called these peaks "C I" and "C II" respectively) were greatly influenced by the retinal location of the stimulus. The polarity of these components reversed when the upper rather than the lower f i e l d of vision was stimulated, and the amplitude of C II changed depending upon the distance of the recording electrode from the calcarine fissure i n both the 11. horizontal and v e r t i c a l planes. Based on these and other observations of responses to right and l e f t half f i e l d stimulation, Jeffreys & Axford concluded that the two peaks had spatially separate sources; the striate cortex was the source of C I and the extra-striate cortex on the outer surface of the o c c i p i t a l lobes was the source of C I I . Halliday & Michael (1970) described a single component at 100 msec, i n the response to a reversing checkerboard stimulus that also showed polarity reversal depending on the part of the visual f i e l d that was stimulated. They suggested that this component originated i n extra-striate cortex i n the o c c i p i t a l pole above and below the calcarine fissure. A similar component i n the responses to a reversing pattern of red lights was described by Purves & Low (1975) using separate stimulation of upper and lower visual f i e l d s . The polarities and latencies of the components described by a l l of these groups are somewhat different probably due i n part to differences i n the luminance of the stimulus and i n the techniques of stimulus presentation. The lack of standardized stimulation and recording techniques makes the findings of the different investigators d i f f i c u l t to compare. Data from intracellular recording i n the visual cortex i n cats and monkeys (Hubel & Weisel, 1965 & 1968); Bishop, Coombs & Henry, 1971) have shown that many c e l l s i n this brain region are much more sensitive to contoured than to unstructured lig h t stLmuli. Such work has resulted i n important advances i n the understanding of the c e l l u l a r physiology of the o c c i p i t a l lobe and of the functional substrates of vision. Data from the study of VER's recorded i n awake human subjects (see the above observation, localizing the origin of these pattern specific response components to 12. striate and extra-striate cortex) i n turn, have provided supportive evidence of the existence of these pattern sensitive mechanisms i n this region and they may prove to be an important tool for further investigation of visual perceptual processes i n man. In addition to the studies concerned with the topographical mapping of the response components, there have been many others that have demonstrated the effect of a variety of patterned stimulus parameters on the evoked co r t i c a l response. Clynes & Kohn (1967) reported one of the earl i e s t studies using stimulus patterns other than the relatively simple ones (checkerboards and stripes) that were used by Sphelmann (1965). They studied the spatial distribution of the evoked potentials to a large number of visual stimuli including complex patterns and colours and found that there were "spatially independent components" that were peculiar to each type of stimulus for a given individual. They also noted that while the latencies of these components were very constant, their amplitudes were quite variable, a finding replicated i n many later studies. The changing polarities and amplitudes of the response components to different stimuli which they reported do suggest the po s s i b i l i t y of changing sources and sinks of current i n the c o r t i c a l a c t i v i t y evoked by the different stimuli. However their rather idiosyncratic electrode placements (in a rosette pattern around the oc c i p i t a l regions), their use of bipolar recording derivations, and their large number of stimulus types makes i t impossible to compare their results to those of any later investigators, or to make any definite interpretations of the meaning of their results. 13. Rietveld, Tordoir, Hagenouw, Lubbers & Spoor (1967) and Harter & White (1968) presented more detailed studies of the responses to a checkerboard pattern and showed that the amplitude of the response from 100 to 200 msec, post-stimulus was sensitive to the size of the checks i n the pattern. They showed that i n contrast to the flash evoked responses i n which the luminance of the stimulus i s an important determinant of the latency of the components, (Creutzfeld & Kuhnt, 1967; Tepas, Guterus & Klingman, 1974), the latency of the components i n the pattern response did not change over a wide range of luminances. They suggested that i t was the presence of the intersecting contrast borders that was essential for the patterned type of evoked c o r t i c a l response. How sharply the stimulus pattern was focused was also found to have a significant effect on the VER amplitude i n these studies. Many subsequent reports (Arden & Lewis, 1973; Harter & White, 1970; John, 1974) have demonstrated that the amplitude of the VER to patterns consisting of different sized checks i s a useful tool i n the determination of the refractive error i n c l i n i c a l ophthalmology. The patient's response i s of largest amplitude for the smallest size of checks on which he can focus clearly. A number of other stimulus parameters have been shown to be reflected i n the VER configuration besides focus, check size and luminance as indicated above. The overall size of the stimulus for both unstructured (Rietveld, Tordoir & Duyff, 1965) and structured (Harter, 1971; Rietveld et a l . , 1967) f i e l d s has been found to be important with a marked difference between foveal and extra-foveal stimulation. The colour of the stimulus can also 14. influence the response (Kinney, McKay, Mensch & Luria, 1972; Clynes & Kohn, 1967; Shipley, Jones & Fry, 1965) as can i t s orientation (Kulikowski, 1974; Yoshida, Iwahara & Nagamura, 1975) although there i s some dispute about the l a t t e r (Kakigi, Miyazaki & Mori, 1972). Responses have been demonstrated both to on- and off-set of diffuse (White & Easton, 1966) and patterned l i g h t (Harter, 1971). The specific findings i n these reports that are relevant to the present study w i l l be considered i n more det a i l i n the Discussion Section. 2.3 Evoked responses i n the study of psychological variables and meaning. A very large number of publications i n the evoked potential literature have originated i n laboratories of physiological or experimental psychology. A review of the relevant works of necessity includes terms and phrases, the precise definitions of which are outside the scope of this thesis. In appendix C a discussion of the terms "psychological variables" and "stimulus meaning" has been included to provide some background for the reader unfamiliar with these concepts, and to c l a r i f y their use i n this text. The concept of stimulus meaning i s also considered i n relationship to the experimental findings of this dissertation i n section 6.5 of the Discussion chapter. A great many reports of the effects of a variety of psychological variables on the evoked potential configuration are reviewed by Regan (1972) and Beck (1975). These studies can be divided into two categories, those that are concerned with the entire evoked response configuration and those that 15. are primarily concerned with a late positive component of the evoked response (also called "P "^ or "P3QQ" because i t appears as a prominent positive peak at approximately 300 msec, after the stimulus). Chapman & Bragdon (1964) i n one of the ea r l i e s t studies of the effect of psychological meaning on the evoked response not s p e c i f i c a l l y concerned with the component, compared evoked responses to task relevant (blank) and irrelevant (patterned) stimuli and found that the responses to task relevant stimuli were larger and different than those to the irrelevant stimuli. However they f a i l e d to consider that there would have been a difference between the responses to their blank and patterned stimuli anyway whether they were task relevant or not. L i f s h i f t z (1966) and Begleiter, Gross & Kissin (1971) i n similar studies claimed that the VER to pleasant, repulsive and neutral categories of visual stimuli were different for each category, and that the effect disappeared when the stimulus was blurred so as to be unrecognizable. John, Herrington & Sutton (1967) i n a widely cited study on meaning found that the VER was different for different geometric forms of equal area and similar for versions of the same geometric form of unequal area. They concluded that this evidence suggested the wave form of the evoked response reflects at least i n part the symbolic meaning of the stimulus. Austt, Buno & Vanzulli (1971) i n a similar study claimed that both the o c c i p i t a l visual evoked response and the visual evoked response recorded at the vertex are related to the processing of sensory information. The o c c i p i t a l response was said to be more specific for visual stimuli than the one from the vertex and was modified by changes i n visual perception, significance of the stimulus and the program the subject was performing. 16. Buchsbaum & Fedio (1969) i n a study of the visual evoked response to different patterns, including some words, also found differences i n the VER waveforms that were related to meaning. Because of their relevance to the findings of the present study these l a t t e r three studies w i l l be considered i n further detail i n the Discussion section, Begleiter & Platz (1969) showed that the response evoked by a particular visual stimulus could be modified by c l a s s i c a l conditioning. A negative peak at 155 msec, to 160 msec, latency was enhanced when the stimulus was conditioned by association with a loud t r a i n of clicks and this enhancement was reversible with extinction and reconditioning. In three more recent reports (Begleiter, Porjesz, Yerre & Kissin, 1973; Porjesz & Begleiter, 1975; Begleiter & Porjesz, 1975) of the effect of the subject's expectancy or perception of a stimulus, these authors have shown that the amplitude of the vertex VER to a flash of medium intensity was higher i f the subject expected or thought he saw a bright flash and significantly smaller i f he expected or thought he saw a dim flash. These findings are reminiscent of John, Shimokocki & Bartlett's (1969) and Ruchkin & John's, (1966) findings of "neural read-out from memory during generalization" i n cats. These investigators trained cats to discriminate between two stimulus f l i c k e r frequencies. They found that the evoked potential i n various regions of the brain to a stimulus of an intermediate f l i c k e r frequency closely resembled the response to the appropriate signal for the behavior which was displayed on that t r i a l ; the configuration of the response evoked by the intermediate stimulus was different depending on whether the cat made the behavioral response associated with the high or low frequency f l i c k e r . 17. The authors concluded that this constituted evidence for release of a neural process representing previous experience and provided some further support for the s t a t i s t i c a l theory of learning and memory suggested by John (1967). Investigators whose work has centred on the experimental manipulation of the late positive (P^) component of evoked potentials have suggested a large number of possible psychological variables that may relate to this peak. I t has been variably proposed that the "P^" i s a physiological sign of "delivery of task-relevant information" (Donchin & Cohen, 1967), of the "resolution of prior uncertainty" (Sutton, Tueting, Zubin & John, 1967) of "cognitive evaluation of stimulus significance" (Ritter & Vaughan, 1969), of "a decision regarding the stimulus" (Rohbaugh, Donchin & ErUcsen, 1974), of a "reactive change i n the state of arousal" (Karlin, 1970), and of "stimulus salience" (Jenness, 1972). Some authors (Karlin, 1970, Naatanen, 1969) have argued a non-cognitive explanation for a l l of these findings, suggesting that the difference i n the responses to the relevant or attended stimuli i n a l l of the experiments can be explained by a "prior-state" hypothesis, meaning that a l l of the findings could be explained on the basis of a change i n the subject's state of arousal or expectancy and not by any specific parameters of the stimulus i t s e l f . However, Picton & Hillyard (1974) and Ford, Roth, Dirks & Kopell (1973) i n more recent work studying evoked response correlates of selective attention have shown that this i s not true. A l l have demonstrated two separate systems operating during selective attention (for auditory stimuli); one indicated by increased amplitude 18. of waves and involving selection of a particular input channel and the second indicated by the more widespread complex and involving complex information processing or decision making based on information provided by the stimulus. Harter & Salmon (1972), i n an e a r l i e r study, had also found a consistent "difference waveform" with an e a r l i e r negative and later positive peak between responses to task relevant and task irrelevant patterned visual stimuli (although the non-specific "prior state" of the subject was not as carefully controlled i n this work as i t was i n the later studies that used auditory stimuli). I t has also been shown that the component i s not e l i c i t e d by a l l attended and task relevant stimuli or emitted stimuli (see below) that require decision, but only by those for which certain requirements of detection confidence and prior subject uncertainty have been met (Squires, Hillyard & Lindsay, 1973; Sutton, Braren, Zubin & John, 1965; Tueting, Sutton & Zubin, 1971). In a recent report on the P^ component during an auditory detection task with cued observation intervals Squires, Squires & Hillyard (1975a & b) have further c l a r i f i e d the effect of these two major factors of decision confidence and expectancy. They suggested that the P^ (because of i t s topography and sensitivity to the probability of the stimulus occurrence or absence) i s the same physiological brain event, whether i t occurs i n a threshold detection task, omitted stimulus paradigm (Picton, Hillyard & Galambos, 1973) or a single-double stimulus discrimLnation task (Sutton et a l . , 1967). A few authors have reported that a c o r t i c a l response can be e l i c i t e d by omission of an expected stimulus; these responses to stimuli that are 19. expected but do not occur have been called "emitted potentials" by Weinberg, Walter & Crow (1970) and this term has been adopted by most other investigators of this phenomenon. Sutton, Ruchkin & Tueting (1974) reviewed most of the relevant studies of the component of the emitted potential and the experimental paradigms used for i t s production. Squires et al.(1975b) provided some further experimental evidence that c l a r i f i e s some of the questions on the appearance of the P^ i n threshold detection tasks raised i n the presentation of Sutton et a l . (1974) and they also demonstrated that the essential c r i t e r i a for a stimulus absence to e l i c i t a P^ wave are that i t i s clearly recognizable and relatively improbable. There has been some controversy about the s p e c i f i c i t y of the emitted response. Sutton et a l . (1974) stated that there was no evidence available indicating that the P^ component of the emitted potential i s specific for the modality or intensity of the stimulus. Ritter, Simson & Vaughan (1974) however i n another paper from the same symposium did provide some experimental evidence that the negative component e l i c i t e d by the missing stimulus was different for the two modalities they used (visual and auditory) but they did not investigate s p e c i f i c i t y within a modality. Other authors who have described emitted responses have been investigating effects of stimulus timing (Klinke, Fruhstorfer & Finkenzeller, 1968) selective attention (Picton et a l . , 1973) or the subject's expectancy i n a guessing paradigm (Weinberg et a l . , 1970) and none of these experiments provides any evidence on the s p e c i f i c i t y of the response to the modality or meaning of the omitted stimulus. 20. 2.4 Summary of the literature review. I t has been established that sensory evoked responses recorded from the scalp r e f l e c t dendritic post-synaptic potential a c t i v i t y i n the underlying c o r t i c a l c e l l s . I t has also been shown that only widely synchronized c o r t i c a l a c t i v i t y appears i n scalp recordings because of the attenuation by the intervening layers of meninges, cerebrospinal f l u i d and s k u l l . The EEG waveform recorded after presentation of a sensory stimulus consists (after the on-going, non-event related EEG a c t i v i t y has been extracted) of a series of well defined peaks or components. Experimental evidence suggests that at least some of these components originate i n different c o r t i c a l regions. I t has also been shown that these different components are sensitive to changes i n a variety of stimulus parameters. The use of patterned visual stimuli to produce evoked responses i s a relatively recent development that allows the study of much more complex aspects of stimulus meaning than did the older method of stroboscopic flash presentation. Components of the pattern VER have been shown to be sensitive to focus, to the number of lines per unit area (contour density) and to a lesser extent to the luminance of the presented stimulus. Steady state evoked responses (related i n principle to transient evoked responses, but e l i c i t e d by a different stimulation technique) to grating patterns of varying contrast levels and spatial frequencies have been shown to be good indicators of the subject's threshold for these patterns as determined by separate psychophysical threshold testing. 21. There i s also a large body of literature concerned less with the physical parameters of the stimulus and more with the effect of psychological variables on the evoked response configuration; both with non-specific variables such as attention, and with more specific cognitive variables such as stimulus meaning or decision-making based on information provided by a stimulus. I t has been shown, for example, that a late positive component of the response i n any modality i s related to this decision making process, but that the wave appears i n the evoked response only i f the subject i s certain he detects the "stimulus" and i f the stimulus occurs rel a t i v e l y infrequently during the paradigm. I t has also been claimed that the evoked response configuration i s affected by the symbolic meaning of the stimulus. However, the few published studies purporting to demonstrate such a relationship have f a i l e d to define clearly either the experimental changes made i n stimulus meaning or the effects they produce on the configuration of the evoked response. There have been a few reports of the recording of c o r t i c a l responses (at the scalp) e l i c i t e d by stimuli that are expected by the subject but do not occur. These have been called emitted potentials since they are not evoked by an external stimulus but presumably r e f l e c t some brain process related to the remembered stimulus. I t has been suggested that this type of response recorded at the scalp might provide a useful method of examiriing brain processes related to stimulus rteaning without cmtamination by a c t i v i t y that i s dir e c t l y related to the sensory input. To this date however, neither the techniques for obtaining emitted potentials, nor the configuration and topography of these waveforms have been well documented i n the literature. 22. Chapter I I I STATEMENT OF THE PROBLEM FOR INVESTIGATION The question of whether any part of the response evoked by patterned stimuli reflects higher perceptual rather than simple sensory process i s s t i l l unresolved. In two recent, comprehensive reviews of the f i e l d (Ciganek, 1975; Regan, 1972), the claims of John et a l . (1967) were singled out as c r u c i a l because of their significance for the study of evoked response correlates of higher processes. Both reviewers urged that these findings require independent validation and development. To this date there has been no report of such studies. Therefore a series of experiments was designed to study the effect of the geometric form and meaning of the stimulus on the configuration of the visual evoked response. More s p e c i f i c a l l y i t was proposed: (1) To establish whether there are reliab l e or consistent differences between evoked responses to geometric shape stimuli and to determine the spatial (scalp distribution) of any differences. (2) To demonstrate the effect on the evoked response of assigning a meaning (as cues i n a task) to the different shapes. (3) To record the "emitted potentials" occurring when a subject expects but does not actually see these shapes, and thus determine i f i t i s possible to record EEG correlates of brain processes related to the meaning of a stimulus i n the absence of the stimulus i t s e l f . 23. (4 ) To apply a variety of analytical techniques to the evoked potentials obtained to explore means of quantifying any difference between the complex waveforms recorded i n the different experimental paradigms. In a l l of the experimental work, data collection and analysis were performed by a d i g i t a l computer where possible. Single t r i a l samples were saved so that'a variable and small number of t r i a l s could be used i n the averages and so that alternate analysis techniques such as a stepwise discriminant analysis, manipulation of single t r i a l s prior to averaging and computation of the ratio s t a t i s t i c called the A descriptor (John et a l . , 1967) could be carried out. (See section 6.2 for a detailed discussion of these analytical techniques and Appendix B for a description of stepwise discrimiiiant analysis. Because i n averages of small numbers of t r i a l s , any a c t i v i t y of rela t i v e l y high voltage present even i n only one t r i a l affects the averaged wave form to a disproportionate extent, an automatic artefact rejection routine was included i n the data acquisition program (see Methods). Chapter IV METHODS 4.1 Data Acquisition Methods. For the f i r s t part (paradigm 1) of the experiment thirteen subjects were used. Ten of these were paid women student volunteers between the ages of 19 and 28. The other three were members of the Laboratory Staff. A l l of the subjects were tested at least once and some were tested repeatedly but always on different days separated by at least a week. A tot a l of 26 runs of paradigm 1 were used for the analysis. Five of the thirteen subjects, including two of the staff members, were used for the second part of the experiments (paradigm 2) , i n which from three to ten separate recording sessions for each subject were carried out. The subjects sat alone i n a shielded room with low intensity overhead lighting (.5 ft-candles) and watched a television screen 1 meter distant for the presentation of the stimulus. The electrodes were Grass gold discs applied with Grass electrode paste according to the International 10-20 system (Jasper, 1958) for electrode placement. (See Fig. 1 for diagramatic representation). For a l l of paradigm 1 the standard locations of 0 2, 0^ P 4, P^, C 4, C 3 and C z were used. In paradigm 2 these locations were also used but i n some experiments 0 , P , C , F and two locations designated CP 0 and CP. located halfway Z Z Z Z j 4 between Cj and P^ and C^ and P^ respectively were used. A l l recording was referred to the contralateral ear or to the l e f t ear i n the case of midline locations. 25. Fig. 1 Diagramatic representation of the position of the electrodes on the scalp. 26. The EEG was recorded on a Beckman Dynograph Type R with a time constant setting of .3 sec and high frequency cut-off f i l t e r at 30 Hz (a setting of 3 on the power amplifier f i l t e r ) giving an effective system band width of 0.5 - 30Hz (3dB points). From the Dynograph the EEG signals on seven channels were led to A-D converters of the computer (a detailed description of the computer system used i n these experiments i s given i n Appendix A) and sampled every 2 msec, for a 512 msec, epoch, beginning 56 msec, before the stimulus presentation. Single t r i a l s (256 points per t r i a l ) were stored on the disc of the PDP 11 computer system and a l l data was transferred to Dectapes for permanent storage at the end of each paradigm. The experimental set-up i s shown i n the diagram i n Fig. 2. The experiments were a l l controlled i n real time by a program called VER03 which ran on the PDP 11 system. The flow chart of the program i s given i n Appendix A with some explanation. An automatic artefact rejection routine was included. This routine checked each t r i a l sample to determine i f i t exceeded some preset amplitude limits, and i f i t did the t r i a l was not saved. The program then continued the experiment u n t i l the required number of t r i a l s had been collected i n each group to be sampled. (The number was 30 t r i a l s of each shape i n paradigm 1, and 10 t r i a l s of each shape i n paradigm 2). I t was found from observing the paper writeout of the EEG a c t i v i t y that this technique effectively excluded any t r i a l s where there was eye blink or subject movement. In some experiments the EGG (electro-oculogram) was recorded on one of the seven channels to check for any event-related slow eye movements which might not have been rejected by the system. D a t a A c q u i s i t i o n S y s t e m . Fig. 2 Experimental set up. The stimuli were the four different geometric shapes shown in Fig. 3. They included two familiar shapes, (one with corners and one without), a square, and a c i r c l e , and two unfamiliar and less easily named shapes called here the e l and the omega. They were a l l shown as white outlines on a black background. A l l had approximately equal length of contrasting border. They were presented as slides by a Kodak Carousel projector. The images from the projector were monitored by a television camera and were transmitted to the T.V. monitor placed one meter i n front of the subject. An electromagnetic shutter (Gerbrant) on the lens of the projector controlled the brief 20 msec, presentation of the stimulus. Advancement of the carousel and the shutter opening were controlled by the computer. The size of the shapes on the T.V. screen was approximately 6 cm. by 6 cm.; therefore the shapes subtended an angle of 3.6 degrees at the centre of the subject's visual f i e l d . The luminance measured at the screen was 10 ft-candles for the background and approximately 25 ft-candles for a l l four stimuli. The subject was instructed to fixate visually on a point on the centre of the screen throughout the recording period. For paradigm 1 (see Figure 4 for a diagramatic representation), the four stimuli were presented i n a random sequence at from 3-5 sec. intervals u n t i l 30 t r i a l s of each shape were collected. The number 30 was chosen because i t was the lowest number per average that produced consistent evoked potential waveforms and was the highest number of single t r i a l s i t would be possible to store i n the computer memory during a single experiment. 29. Fig. 3 Stiraulijs Shapes 1.circle 2.square 3.omega 4.el Stimulus presentation 20 msec fc==l-*H 56 msec *<— Interstimulus interval - random 3 - 5 sec 512 msec Sampling interval Shutter stays closed O L? 4- • A O 512 t msec Button press I——1.4 sec-^i Regular interstimulus intervals or buzzer sounds Fig. 4 Diagramatic representation of the experimental paradigms. The upper part of the figure shows the taming of the stimulus presentation i n paradigm 1, and the lower part shows the same for paradigm 2. 31. The subjects were instructed to watch the presentation of the different shapes and to try to stay aler t and attentive. Total time for paradigm 1, including one or two brief interruptions, was from 10 to 15 minutes. The ongoing EEG was monitored on the paper writeout by the experimenter and when the Alpha rhythm began to dominate the record or any Theta a c t i v i t y appeared, the experiment was interrupted and the subject asked to refocus her attention after a brief rest. In paradigm 2 (see Figure 4 for a diagramatic representation), five shapes, a c i r c l e , e l , cross, square and omega were presented i n the same order at regular intervals of 1.4 sec. In approximately one third of the sequences either the square or the e l was omitted (the shutter did not open). The experiment continued u n t i l the EEG during ten omissions of each shape was collected and stored. With no rejections due to artefact this required 63 sequences of five shapes, or 315 stimuli, which took lh minutes. Ten responses to each of the four shapes used i n paradigm 1 when they did appear were also saved. The subject was instructed to learn the sequence of five shapes, then was told that occasionally the square or e l would f a i l to appear. She was given two buttons (hand held thumb-switches) and instructed that i f the square f a i l e d to appear at the expected time, she must press the left-hand button and when the e l was emitted she must press the right-hand button. Failure to respond correctly within 400 msec, of when the stimulus should have appeared i n the rhythmic sequence, or pressing the button when the stimulus did appear resulted i n a warning tone being sounded (controlled by the computer). The rhythmic presentation sequence was not broken by the warning tone. Button presses were recorded as square wave pulses on one of the seven data channels so that the exact timing of the response could be compared to the EEG ac t i v i t y . I t also provided a way of counting the number of correct responses at the end of the experiment. The subjects were allowed to practise this f a i r l y d i f f i c u l t task prior to the recording session and a l l became quite proficient i n making the correct response within 400 msec, in from 70 - 100% of the t r i a l s with the missing stimulus, and i n making very few responses when the stimulus was present. At the end of an experimental session the data were averaged, 30 t r i a l s per average for paradigm 1, and 10 t r i a l s per average for paradigm 2 and the averages were immediately written out. A l l of the data (as single t r i a l s ) was then transferred to Dectapes for permanent storage and could be accessed from there by a program called GASP for further analysis. Q Calibration of the data acquisition system was done at the beginning of each experiment by inserting a 5 Hz. 50 yv square wave signal from a Grass calibrator into one pair of inputs at the headboard. A l l seven recording channels were calibrated using this pair of electrodes. The VER program was then run and a writeout of some single t r i a l s made. The gains were adjusted on the EEG amplifiers so that the 50 yv signal i n a single t r i a l appeared with an amplitude of 25 mm. on the plotter. Since a l l averaged EEG signals on the f i n a l plot were true means (rather than sums) a 50 yv signal appeared with a 25 mm. amplitude i n the f i n a l plot whether i t was the signal of a single t r i a l or the signal i n the averaged response. 4.2 Data Analysis Methods. The differences between the evoked potential waveforms e l i c i t e d by the different shapes were analyzed by a variety of techniques. They include visual inspection for identification and measurement of the prominent peaks of the waveforms and two more complex quantitative approaches. The simpler of the two i s the computation of a descriptor s t a t i s t i c called X that was described by John et a l . (1967). This s t a t i s t i c gives a quantitative measure of how different two waveforms are and the formula and methods used to compute i t are given below. The second method of determining the differences between the responses i s called discriminant analysis, a technique which i s based on multivariate s t a t i s t i c a l theory. A description of the technique and the assumptions on which i t i s based i s given i n Appendix B. The discriminant analysis performed on this experimental data was carried out by the BMD:07M program i n the UBC computing center. The input data for the program consists of two or more groups of observations (in this experimental work each "group" was a shape and each observation was the EEG a c t i v i t y recorded during a single presentation of that shape). Each observation i n turn consists of a number of variables which i n this study were the digitized values of the EEG voltage at different time points. The program proceeds to choose a subset of the variables (i.e. voltages at different time points) and calculates a discriminant function using these variables that w i l l discriminate between the different groups that were entered into the program. This mathematical function can be evaluated for 34. a new observation (a single EEG sample) and i t w i l l provide a prediction of which group (shape) this new observation was most l i k e l y to have come from. The method the program uses to select the variables and to compute this discriminant function i s described i n Appendix B. Certain parameters for this process must be set at the time the data i s entered and these are also given i n that appendix. Another related, but much simpler program called UBC:CLASS evaluates the discriminant function for new data. In order to submit the EEG data to the UCLA BMD:07M program (run on the University IBM 370 computer) the selected digitized t r i a l s (one channel at a time) were transferred from Dectape to either paper tape or magnetic tape on the PDP 11 system. During the transfer the digitized t r i a l representation was reduced from 256 points to 67 points by omitting the f i r s t 26 points (which represented the 52 msecs. of the pre-stimulus baseline), then taking the average of each of the 67 sets of 3 points that followed and omitting the l a s t 29 points. The data from one channel at a time were then transferred to f i l e s on the University IBM 370 system. Each 67 point single t r i a l representation included a 5 character label indicating which shape i t came from. The BMD:07M program was run for either 2 or 4 groups (each group consisting of 30 t r i a l s e l i c i t e d by a different shape). On a few runs 60 t r i a l s per group were included but the separation of the groups by the program did not appear to be improved by this increase i n the number of observations per group. The number of variables per observation was 67 as indicated i n the previous paragraph and the program selected 6 or 7 of these (i.e. 6 or 7 latency points) as the basis of the discriminant function. 35. P i l o t studies showed that allowing the program to run through more than 6 or 7 steps (i.e. to choose more variables than this) did not improve the discrimination and so the program was instructed to stop at this number of steps. This discriminant function produced by each run of the program was stored i n a separate f i l e and the names of the variables i t selected were recorded separately (e.g. the 5th, 21st, 35th, 42nd, 51st and 60th variable).. For testing one of these functions, data from a l l of the experiments for the same subject for the same channel that had not been used for input to the BMD:07M program that determined the function, were put i n a single f i l e . The program UBC:Class was run on this f i l e with the discrhntlnant function and a format statement indicating which were the six or seven variables (latency points) to be selected from the response for this particular function as i t s input. The results of this testing were expressed as percentages of correctly c l a s s i f i e d t r i a l s and i n the case of the 4 group choices, the wrongly c l a s s i f i e d t r i a l s were also grouped as to which shape they were mistaken for and these percentages were tabulated. To compute the X ratios as described by John et a l . (1967) averages of 10 t r i a l s were prepared on the PDP-11 system with the GASP program. These were transferred by magnetic tape to the IBM 370 where a Fortran program computed the X values for the 4 averages given i t from the following formula: = ( d l , 3 + d2,4 } / ( d l , 2 + d3,4 ) where d.. i s the absolute value of the root mean square (r.m.s.) difference between waveforms i and j . /"r.m.s. = x. /n where x. i s the i sample 36. of the waveform with the value of the baseline as computed from the f i r s t 3 points (18 msec.) of the response subtracted from i t V The subscripts 1 and 2 denote the two replicated responses to one shape and 3 and 4 denote the replicated responses to the second shape. Analysis of the responses to the omitted stimuli was made somewhat d i f f i c u l t by the relatively small number of these responses available. The square or e l was omitted from i t s position i n the sequence only occasionally. An experiment consisting of presentation of at least 320 stimuli provided only 20 "emitted" responses, ten each for the square and e l . In the averages of only ten t r i a l s the background EEG (noise) level was sometimes high enough to make i t d i f f i c u l t to visually separate the components of event-related ac t i v i t y , especially i f these were rel a t i v e l y small i n amplitude. However, combining data from more than one experiment to provide more responses for averaging introduced added v a r i a b i l i t y to the responses because of changes i n the subject's performance level , state of alertness or i n technical elements such as minor differences i n electrode placement. In view of these considerations, for presentation of the results, the measurements made by visual inspection for compiling tables VTI and VTII were made on averages of ten t r i a l s from each single experiment, but for the ill u s t r a t i o n s of the responses 30 t r i a l s per average were used so that the evoked response components could be more easily visualized. After the conventional averages were produced and examined, an alternate technique similar to the one used by Ruchkin (1974) was tr i e d to compensate for the effect of a varying estimate of the time of occurrence of the 37. omitted stimulus by the subject. Its purpose was to have a l l of the peaks from the single t r i a l s coincide i n time prior to averaging, so that any ac t i v i t y that was time-locked to the P^ peaks would be enhanced. The single t r i a l s at C were f i r s t f i l t e r e d by a 0 - 6 Hz band-pass d i g i t a l f i l t e r to make the peak of the slow positive a c t i v i t y clearer for measurement. The f i l t e r e d t r i a l s were then moved to a display screen. A clear positive wave could be seen i n about 90% of the responses examined, between 240 and 450 msec, after the stimulus should have occurred. (If no peak could be seen i n that interval the t r i a l was not included for the "shifted" average). The latency of the most positive peak was measured and the size of the s h i f t along the time axis (in msec.) that was required to move the peak of the largest positive wave to an a r b i t r a r i l y chosen central point (350 msec.) was calculated. As a simple control, this shifting procedure was also applied to some responses from single t r i a l s e l i c i t e d by the square and e l when they were present. This i s further discussed i n section 5.3.2. The GASP program mentioned included routines for retrieving the single t r i a l s from any experiment from Dectapes, averaging any number of responses, switching around the seven channels i f necessary i n cases where different channels had been used for a given scalp location, displaying averages or single t r i a l s on the display screen and for plotting any of these responses before or after manipulation on a printer-plotter. The program also had the capacity for d i g i t a l f i l t e r i n g of data through whatever bandpass was required. 38. Chapter V RESULTS 5.1 Effect of the Different Experimental Conditions on the Evoked Response. Before examining the effect of the stimulus shape on the configuration of the evoked response, i t i s necessary to assess the contribution of the different presentation conditions of the two experimental paradigms i n order to decide whether the data from the two could be combined for r subsequent analysis. In the stimulus sequence of paradigm 2, the c i r c l e and omega were always present and served only as cue stimuli. For the square and e l , on the other hand, since the appearance of the appropriate shape meant that no button press was to be made (detected omissions were signalled by pressing one of the two switches) and since there was some uncertainty as to whether the square or e l would appear i n i t s time interval i n any given stimulus sequence, the presence of these stimulus shapes was informational!/ more important for the subject. In the evoked potentials e l i c i t e d by the c i r c l e and omega there was essentially no difference between the responses recorded i n the two paradigms at any of the five recording sites common to both experiments. Figure 5 shows these responses for one subject i n the two paradigms and i t can be seen that the evoked potentials from paradigm 1 and 2 are very similar. 39. 100 msec Fig. 5 Effe c t of experimental conditions for the c i r c l e and omega. Paradigm 1 (no task) responses are shown by s o l i d lines and paradigm 2 (task) responses are shown by dotted lines. n=30 trials/average. (Subject SC Exp. 18A,34,35) In the evoked potentials e l i c i t e d by the square and e l however, there were consistent differences between the two paradigms, but only i n the components later than 220 msec. In paradigm 2, but not i n paradigm 1, a broad positive component with a maximum amplitude of 8 to 18 pv (depending on the subject) between 220 and 320 msec, post-stimulus was seen i n the responses e l i c i t e d by the square and e l shapes. This difference i s i l l u s t r a t e d i n Figure 6 for one of the subjects. This late positive wave appearing i n paradigm 2 was largest at C , smaller i n amplitude but definitely present i n the right and l e f t parietal regions, and small and inconsistently present i n the o c c i p i t a l regions. For a given subject this positive wave was of the same amplitude and configuration for both the square and e l responses. There were no amplitude asymmetries between and C^, CP^ a n a- CP^ (not shown i n Fig. 6) and P^ and P^. In summary, the effect of the task imposed i n paradigm 2 on the evoked responses to the four present shapes was as follows: 1. The effect was to introduce a broad positive wave in the evoked responses at C , which was only variably present and of smaller 2 amplitude in o c c i p i t a l regions. 2. This wave was only seen i n the responses e l i c i t e d by the square and e l shapes and not to the c i r c l e and cmega shapes. 3. I t was seen only i n components 220 msec, or later post-stimulus. 41. + 100 msec F i g . 6 E f f e c t o f e x p e r i m e n t a l c c n d i t i o n s f o r the square and e l . Paradigm 1 (no task) responses a r e shown by d o t t e d l i n e s and paradigm 2 (task) responses a r e shown by s o l i d l i n e s . n=30 t r i a l s / a v e r a g e . (Subject MP, Exp. 7,53,54) In view of the above findings, i n considering the data for analysis of the effect of the different shapes on the evoked responses, the c i r c l e and onega t r i a l s from paradigm 1 and 2 were mixed. For the square and e l shapes there was a difference between the paradigms which was maximal at C and started only 220 msec, post-stimulus. Since i t was very small i n o c c i p i t a l regions the results of computerized analysis of the different responses at the o c c i p i t a l locations for these two paradigms were also combined for consideration i n the section to follow. 5.2 Effect of different stimulus shapes on the evoked responses. Consistent differences between the evoked responses to the four different shapes were observed. In order to best define and quantify these differences, three techniques were used. F i r s t , the data from the standard paradigm 1 condition (no task) was examined visually, the components labelled and amplitude and latency measurements made as described below. Secondly, the digi t i z e d single t r i a l s from o c c i p i t a l channels of one experiment were submitted to the BMD:07M Stepwise Discariminant Analysis Program; a function to discriminate between shapes for each subject was computed and then tested on large numbers of single t r i a l s from other experiments with this subject. Finally, the A s t a t i s t i c as described by John et a l . (1967) was computed using averages of ten t r i a l s for most of the experiments. 5.2.1 Results of visual analysis. For visual analysis the evoked responses from thirteen subjects (who had each been run i n paradigm 1 at least once with o c c i p i t a l , parietal, central, and vertex electrodes) were used. Some of the subjects were tested up to five times i n this paradigm but the average number of runs per subject was two. Thirty t r i a l s for each stimulus shape were averaged to obtain the VER's for examination. With subjects that were tested more than once the responses were found to be nearly identical from one experiment to another. A typical set of responses to the four shapes i n seven channels for one subject i s shown i n Fig. 7. The responses showed some variation i n latency and configuration between subjects, but did exhibit essentially the same series of components for a l l subjects. Fig. 8 shows the response at 0^ to the e l shape for a l l thirteen subjects. In a few subjects mid (T^ or T^) or posterior (Tj- or T^) temporal or frontal (F^, F^ or F^) electrodes were used instead of the right and l e f t centrals. No clear response was seen at mid-temporal or frontal locations. The response at the posterior temporals appeared very similar to that seen at 0, and 0„. 100 msec Fig. 7 Evoked responses e l i c i t e d by the four different stirnulus shapes at the 7 electrode locations used i n paradigm 1. n=30 trials/average. (Subject SC, Exp. 1) 45. • LP 100 msec Fig. 8 Intersubject v a r i a b i l i t y . Response recorded from ^1 - A2 e l i c : i - t e d DY t b e e l shape for each of the 13 subjects. n=30 trials/average. For further description of the responses the labelling shown i n Fig. 7 was performed. The f i r s t major positive peak seen at about 95 msec., i n most subjects, was called P p This was sometimes preceded by a definite small negative peak called N^. Following P^ the next large negative peak (which was sometimes double i n some channels) was called N,,, the broader positive wave following N 2 was called P 2, and a later positive peak (when i t occurred) was called P^. I t was noted that the relative amplitudes of these components varied i n the responses to different shapes and among subjects, but i t was usually possible to identify them i n a l l of the recorded channels for a l l subjects. For a l l subjects the amplitude of the P 1 component was approximately equal i n o c c i p i t a l and parietal recordings and sli g h t l y smaller than this i n the central leads. The N 2 component i n six of the thirteen subjects was largest i n the occ i p i t a l regions and either reversed polarity completely or became biphasic (i.e. part of i t reversed i n polarity) anteriorly. The reversal took place either just anterior to or posterior to the parietal electrodes (as i t does i n Fig. 9). In the remaining seven subjects this component became progressively smaller anterior to the oc c i p i t a l locations rather than reversing i n polarity, (as i t does i n Fig. 7). The ?2 conponent had a broad configuration and the latency of this peak was more variable between the different recording sites than was the latency of the other components. In seven of thirteen subjects this wave was of highest amplitude i n the o c c i p i t a l and parietal regions, and i n six of thirteen i t was largest i n the central or vertex channels. Because the components were most clearly defined i n the o c c i p i t a l regions the right and l e f t o c c i p i t a l channels were used to measure the latency and amplitude of the three components for a l l of the subjects. The averages of th i r t y t r i a l s were used for these measurements, although i n some subjects i t was noted that the average of just ten t r i a l s showed the components equally well. It was noted that the latency of the three components for a particular subject was quite consistent for a l l four shapes. Table I shows the mean latencies of the three components for the thirteen subjects' average responses to the four different shapes. It can be seen by inspection of the magnitude of the standard deviations of the mean values i n this Table that there were no significant differences i n the latency of these three components between the different shapes. The mean latency for a l l four shapes and a l l thirteen subjects for P-^  was 95 msec, for N 2 was 148 msec, and for P~ was 219 msec. 48. TABLE I: Latencies of the peaks of the three principal components of the o c c i p i t a l evoked responses to the four shapes presented i n paradigm 1. Each value (in msec.) i s the mean and standard deviation for 13 subjects. N. 2 square 95 - 17 145 - 20 212 - 25 e l 92 - 20 150 - 21 225 - 18 c i r c l e 101 - 23 150 - 26 219 - 21 omega 93 - 24 148 - 25 221 - 24 Mean of 4 shapes 95 148 219 49. There were no consistent differences i n the latency of the components for any of the shapes between right and l e f t o c c i p i t a l regions. I f any small difference between right and l e f t was seen for a given subject, the mean of the two was taken as the value used for preparing Table I. The amplitude of these three components was measured from the pre-stimulus base l i n e . To compare the amplitude of the acti v i t y at 0-^  and 0^ the subjects were divided according to handedness. There were eight right-handers and four left-handers (data from one subject were omitted because the EEG from 0^ was technically inadequate). A mean amplitude difference for each of the four shapes was computed for each subject. For the right-handed group no mean differences between 0-^  and 0 2 amplitudes of greater than 1 yv were seen for any of the components and these were not even consistently i n the same direction. For the four left-handed subjects there was a consistent tendency for N 2 to be sl i g h t l y higher over the right hemisphere (mean difference of 2 uv) for a l l four of the shapes. The mean amplitude of each of the three components recorded at for the four shapes for 12 of the subjects i s shown i n Table II (data from one subject had to be discarded because of technical problems). The amplitudes of the N 2 component of the square vs. e l , c i r c l e vs. omega and c i r c l e vs. e l and of the P 2 component of the square vs. e l , c i r c l e vs. omega and square vs. omega were a l l found to be significantly different using the Scheffe's (1959) method of testing multiple TABLE I I : Amplitudes of the three principal components of the responses at 0 2 to the four shapes presented i n paradigm 1. Each value (in yv) i s the mean and standard deviation for 12 subjects.* P l N2 P2 square 2.9 ±2.4 6.4 ±4.0 6.2 ±3.1 c i r c l e 3.4 -2.4 5.2 ± 2.8 7.2 ±4.5 cmega 3.8 ±3.1 8.1 ± 3.8 10.5 ±5.5 e l 4.8 ±3.4 10.4 ±4.8 9.5 ± 5.0 By Scheffe's (1959) method of testing contrasts (as provided i n the UBC Computing Center Program MFAV) , the amplitudes of the components of the following pairs of shapes were significantly different (p = .01): for the N„ component the square and e l , the c i r c l e and  omega and the c i r c l e and e l were a l l different, and for the P~ component the square and e l , the c i r c l e and omega and the square and omega pairs were a l l different. For the P^ component none of the 6 possible pairs of shapes were significantly different. 51. contrasts (p = .01). There were no significant differences (p = .01) i n the amplitude of the component. I t i s noted that the Scheffe's method i s based on a two way analysis of variance and thus i t i s essentially testing the difference between the shapes for each of the 12 subjects. For two subjects, paradigm 1 was repeated with the T.V. camera lens adjusted so that the shapes appeared on the viewing screen at approximately one half their usual size. The evoked responses to these smaller stimulus shapes were not different (by visual inspection) from the ones evoked by the standard sized shapes. The responses to the small and large square, e l , c i r c l e and omega for one subject are shown i n Fig. 9. 5.2.2 Results of Stepwise Discrirninant Analysis. In a p i l o t study with the BMD:07M program a sampling of data from a l l five subjects from a l l of the recording locations (occipitals, parietals and centrals) was t r i e d to determine which ones would be best for ccnplete testing. Using the U s t a t i s t i c of the program as an indicator of how well the groups were separated (see, Appendix B) i t was found that one of the oc c i p i t a l locations always had the lowest U value ( i . e . showed the best separation) for any pair of groups from a given experiment. 52. Fig. 9 Effect of stimulus size. Responses to standard sized stimuli (visual angle of 3.6°) are shown as s o l i d lines, those to the half-sized ones as dotted lines, (subject ML, Exp. 8,21) n=30 trials/average Therefore, for testing the a b i l i t y of discriminant analysis to cl a s s i f y new data for each subject, only the oc c i p i t a l data were completely tested and are reported here. The work with the BMD:07M and the testing of the functions produced by i t (with the program UBC:CLASS) was done i n two phases. In the f i r s t phase, data collected from four subjects i n paradigm 1 were used. Single t r i a l s from a l l four stimulus shapes were submitted to the BMD:07M program at once, hence the resulting discriminant functions were intended to class i f y any single t r i a l (or observation) as belonging to one of the four possible groups (shapes). In the second phase the data from only two stimulus shapes at a time were submitted to the program. In Table III the results of testing the discriminant functions classifying a l l four shapes for the four subjects on data from right and l e f t o c c i p i t a l locations are presented. The percentages given i n the Table are the proportions of the tot a l number of t r i a l s the function correctly c l a s s i f i e d . The classifications made on the t r i a l s used to compute the function (i.e. a p o s t e r i o r i c l a s s i f i c a t i o n ) were included i n the percentages and tot a l numbers of t r i a l s given i n this Table. 120 t r i a l s , 30 of each shape, were used to compute each function. It can be seen from the Table that the functions c l a s s i f i e d from 35 to 44% of the waveforms from the single t r i a l s correctly, proportions which are significantly greater than the level of 25% expected with a random cla s s i f i c a t i o n (z =3.58 for 35% which i s greater than z n 1 = 2.57). TABLE III: Performance of the "Four shape" Discriminant Functions. Percentage of t r i a l s correctly c l a s s i f i e d by the BMD:07M Stepwise Discriminant Analysis technique on data from right and l e f t o c c i p i t a l locations. Total number of t r i a l s c l a s s i f i e d for each subject i s given i n parentheses. A l l percentages given are significantly* greater than the 25% level expected with a random c l a s s i f i c a t i o n (p = .01). For 35% z = 3.58 which i s greater than z m = 2.57. Subject 0 2 (right) 0-j^  (left) MP (260) 41% 44% SC (480) 37% 37% VS (240) 39% 40% ML (360) 35% 41% mean of 4 subjects 38% 40% * significance determined by evaluating the test s t a t i s t i c z = p - p //pq/n- where p i s the observed proportion, p i s the expected proportion, q = 1 - p and n i s the sample size, z i s considered to be normally distributed. (Bahn, 1972). TABLE III: Performance of the "Four shape" Discriminant Functions. Percentage of t r i a l s correctly c l a s s i f i e d by the BMD:07M Stepwise Discriminant Analysis technique on data from right and l e f t o c c i p i t a l locations. Total number of t r i a l s c l a s s i f i e d for each subject i s given i n parentheses. Subject 0 2 (right) O-^left) MP (260) 41% 44% SC (480) 37% 37% .VS (240) 39% 40% ML (360) 35% 41% mean of 4 subjects 38% 40% 55. To ascertain which of the four shapes the functions were best able to identify, the pattern of the incorrect classifications was studied (i.e. when the response from one shape was c l a s s i f i e d as belonging to the group from another shape). A count was then made of how many incorrect predictions f e l l into each of the six possible stimulus pairings (e.g. how many times a square's response was c l a s s i f i e d as a ci r c l e ' s response and vice versa). In Table IV, the number for each of the six pairs i s given as a percentage of the total number of t r i a l s . I t can be seen from this Table that with this c l a s s i f i c a t i o n technique, i t was the c i r c l e and square and the e l and cmega pairings that were most frequently confused with each other. As described i n the Methods section the BMD:07M essentially chooses a i subset of variables (latency points) on which to base the discriminant functions. The distribution of these latency points chosen for the eight functions (2 for each of 4 subjects) whose performance i s described i n Table III i s shown i n Fig. 10. The dark part of the bars shows where the points chosen i n the f i r s t or second step of the computation of the discriminant function were located. I t can be seen that the highest number of variables chosen were i n the 150 - 180 msec, and 210 - 300 msec, ranges. None of the f i r s t two points were found before 120 msec. This indicates that the major differences between the evoked responses e l i c i t e d by the different shapes were found i n the waveforms between 150 to 180 and 210 to 300 msec, after the stimulus. TABLE IV: Percentage of t r i a l s incorrectly c l a s s i f i e d separately into pairs of shapes confused with each other. Same data as that presented i n Table III. subject circle-square circle-omega c i r c l e - e l MP 17% 7% 5% SC 14% 9% 7% VS 16% 16% 14% ML 14% 4% 10% mean for 4 subjects 15.2% 9% 9% square-omega square-el omega-el MP 7% 7% 14% SC 9% 12% 13% VS 2% 7% 5% ML 6% 12% 18% mean for 4 subjects 6% 9.5% 12.5% 57. 1 0 - 1 0 3 0 6 0 9 0 120 150 180 210 2 4 0 2 7 0 3 0 0 3 3 0 3 6 0 3 9 0 msec after stimulus Fig. 10 Distribution of the variables chosen by the 4-group discriminant functions along the time axis. (Stimulus occurs at 0 msec.) The upper line of each bar indicates the total number of variables within each latency range. The shaded portion indicates the number of the variables chosen i n the f i r s t two steps of the BMD:07M program. In the second phase of the application of discriminant analysis to the evoked potential data, i t was decided to investigate differences between just two shapes at a time, and then to test the discriminant functions on larger numbers of single t r i a l s than i n the f i r s t phase. The BMD:07M program was run for just two groups of data at a time (i.e. the responses from only two shapes were submitted) and a few runs were made for some of the six possible pairs of shapes. The results of these different two-group functions for one subject are presented i n Table V. I t was found that the discriminant functions for the square-el and circle-omega pairs always had the lowest U values of any of the six pairs and also had the highest percentage of correction classifications of new data (as shown in Table V). Given the limitation on computer time, i t was not possible to compute and test a l l of the functions for a l l six pairs for a l l of the subjects. I t was therefore decided to select the two pairs which consistently had the lowest U values i n the p i l o t studies. Then, for five subjects, seme of the responses e l i c i t e d by the c i r c l e and omega and some for the square and e l recorded at 0-^  and were submitted to the BMD:07M; discriminant functions were computed for each pair and these were tested extensively on more data from that subject. (Three of the subjects were the same ones used for the computations using a l l four shapes simultaneously). The data for testing included both t r i a l s from additional runs of paradigm 1 that had not been used i n the f i r s t phase as well as some single t r i a l responses e l i c i t e d by the present shapes of paradigm 2. TABLE V: Performance of the "two-shape" discriminant functions for five of the six possible pairs of shapes for one subject (MP-data from 0^ ) Number of correct predictions made of single t r i a l s on the data used to make the function and on new data. Number of new t r i a l s c l a s s i f i e d given i n brackets. A l l percentages given of correct classifications are significantly higher than 50% (p = .01) + Square vs. omega Square vs. c i r c l e Circle vs. e l Square vs. e l Circle vs. omega NEW DATA 88 67% (290) 70 59% (290) 82 66% (230) 87 76% (320) 80 71% (380) * values also presented i n Table VT. + significance determined by the same method as i n Table III. z = 3.07 for 59% (the lowest value) which i s greater than z n l = 2.57. 60. The results of the two-group functions computed and tested on the larger data base are presented i n Table VT. In this Table the classifications made a posteriori and on new data are presented separately. Only one of the twenty functions ( i t i s marked by an asterisk) did not perform significantly better than the chance 50% level i n classifying new data from the same subject. The c l a s s i f i c a t i o n of the square and e l shapes was sl i g h t l y better than for the circle-omega pair but the difference was not s t a t i s t i c a l l y significant. There was no significant difference between the performance of functions from data from the right or l e f t hemispheres. The distributions of the latency points chosen for the discriminant functions described i n Table VT are shown i n Fig. 11. Most of the points lay between 120 and 270 msec, and none of the points chosen i n the f i r s t or second step of the computation of the functions was ea r l i e r than 90 msec. There was a tendency for the points to be clustered around the 120 - 150 msec, and 180 - 240 msec, regions of the responses, which corresponded to the latency of the N 2 and P 2 components noted i n the visual assessment of the configuration of the responses (Table I ) . 5.2.3 Results of the X Descriptor Computations. X values for pairs of different shapes were computed for each subject as described i n the Methods section.- Table VTI presents the mean value of this X s t a t i s t i c for the five subjects for the square-el and circle-omega pairs at 0-^  and 02- Each value i n this Table i s the mean of four values computed for that subject, location and pair of shapes. A l l four averages 61. TABLE VI: Performance of the "two-shape" discriminant functions. Number of correct predictions made of single t r i a l s on the data used to make the function (a p o s t e r i o r i ) and on new data. Number of new t r i a l s c l a s s i f i e d given i n brackets. A l l functions were made with 60 t r i a l s . A l l percentages given of correct classifications are significantly higher than 50% (p = .05) except the one marked *. Significance determined by the method described i n Table III footnote, z = 2.13 for 59% (n = 140) which i s greater than z = 1.96. Square-el functions: °2 Data °1 Data Subject a p o s t e r i o r i new data a p o s t e r i o r i new data MP 75% 65% (240) 87% 76% (320) SC 72% 63% (240) 72% 65% (240) ML 83% 62% (240) 75% 70% (440) MD 87% 71% (190) 87% 62% (190) SP 83% 58% (140)* 93% 68% (140) mean of subjects: 80% 64% 83% 68% Circle-cmega functions: MP 80% 61% (380) 80% 71% (380) SC 77%. 65% (190) 77% 63% (190) ML 82% 65% (320) 82% 60% (320) MD 93% 65% (220) 78% 60% (220) SP 70% 59% (140) 72% 61% (140) mean of subjects: 80% 63% 78% 63% 62. • CP A T o n A T o 1 0 6 0 120 180 2 4 0 3 0 0 3 6 0 msec A F T E R S T I M U L U S Fig. 11 Distribution of the variables chosen by the 2-group discriminant functions for each pair of shapes at 0^ and along the time axis. The upper line of each bar indicates the t o t a l number of variables within each latency range. The shaded portion indicates the number of the variables chosen i n the f i r s t two steps of the BMD:07M program. TABLE VTI: Mean X values for each subject. Each value i n the Table i s the mean X for 4 pairs of averages of 10 t r i a l s each of the 50 to 350 msec, interval following the stimulus. Square-el comparison Circle-omega comparison location °2 °1 °2 °1 Subject SP 2.4 5.4 1.7 2.2 ML 2.8 3.5 2.4 1.0 MP, 3.5 1.8 2.6 1.8 SC 1.7 1.5 1.0 1.2 MDL 3.6 1.2 2.1 1.9 5 subject mean 2.8±3.2 2.1-3.2 1.9±1.5 1.6-.9 used to conpute one A value were taken from the same paradigm (1 or 2) and from data obtained i n a single experimental session. A l l values i n the Table are based on the 50 - 350 msec, post-stimulus interval of the averages. The mean value of A for the square-el comparison was 2.7 and for the circle-omega pair was 1.7 for the five subjects. The fact that the mean value of this ratio i s greater than one indicates that the evoked potential waveforms of the square and e l and of the c i r c l e and omega are different and this s t a t i s t i c provides a quantitative index of the difference. (See the Discussion chapter, section 6.2.2 for further evaluation of the significance of this measured difference). The distribution of the 40 A values used to prepare Table VTI i s given i n Fig. 12. I t can be seen that the values were skewed to the right of 1.0 and that although the mean A values were greater than 1, many of the individual values were exactly 1 or less. The A values were also computed for the 0-350 msec, and 120 - 300 msec, intervals. The values for the other intervals were not very different from the ones given. The mean of a l l values for the largest 0 - 350 msec, interval was 2.1, for the 50 - 350 msec, interval reported i n Table VTI 2.3, and for a smaller 120 - 300 msec, interval 2.2. 5.2.4 Summary of the effect of the different stimulus shapes on the evoked responses. To summarize this section, the visual evoked response was found to be different for different stimulus shapes. Various measurements of these differences were made on the oc c i p i t a l l y recorded data and the results were as follows: 65. Fig. 12 Distribution of the individual X values for the square-el and circle-omega pairs that were used to compute the mean values presented i n Table VTI. 66. (1) By visual assessment of components and measurements of amplitudes and latencies, the N 2 and P 2 amplitudes were found to be significantly higher for the e l and omega shapes compared to those of the square and c i r c l e . The amplitudes of the component and the latencies of a l l three of these components (P^, N 2, P 2) were not significantly different between the shapes. (2) Using the BMD:07M Stepwise Discriminant Analysis computer program functions were computed to discriminate between responses to a l l four shapes and between those to two shapes at a time, and these were tested on new data. The four-shape functions c l a s s i f i e d 35 - 44% of the t r i a l s correctly (including both a p o s t e r i o r i and new data). The. two-shape functions c l a s s i f i e d approximately 65% (the mean value for a l l 20 functions) of the new data correctly. The latency points chosen by these discrinunant functions corresponded f a i r l y closely to the N 2 and P 2 component latencies which showed significant differences i n visual assessments. (3) X s t a t i s t i c s as used by John et a l . (1967) were also computed for much of the data. The mean X values were 1.7 and 2.7, which indicated that there were significant differences i n the evoked responses to the different shapes. However, these mean values showed large variances and many individual values of X were less than 1. 5.3 Emitted responses (from paradigm 2) 5.3.1 Averaged emitted responses In the averaged responses e l i c i t e d by the omitted squares and e l s , a consistent late wave (mean latency 337 msec, for the 5 subjects) was seen which was of maximum amplitude at C . (See Fig. 13) . Some smaller z and more variable peaks could also be seen between 80 and 180 msec, after the stimulus would have occurred. These components were not always easy to separate from the background EEG, especially i n the averages of only ten t r i a l s , and their latencies were somewhat variable i n the different subjects. One of the more consistently identifiable components was a negative peak between 100 and 180 msec, which was of maximum amplitude at C z and sometimes was clearly seen i n central and parietal regions as well without any consistent asymmetries. The responses at six recording sites are shown in Fig. 14 for one subject. The amplitude and latency of the P^ wave were measured i n the responses to both present and absent stimuli for each experiment. (All measurements were made on averages of ten t r i a l s ) . The mean and standard deviation of the amplitude and latency of P^ are shown i n Table VIII for the five subjects for both the present and omitted stimuli. No difference between the responses to square and els was noted so they were not separated for preparation of this Table. Table IX presents the values of the amplitude and latency for the e a r l i e r negative component (called N2) for both evoked and emitted responses for the five subjects. 100 msec Fig. 13 Averaged emitted potentials recorded from C for each of the 5 subjects. Responses to the omitted" squares and to the omitted els are shown superimposed. n=30 trials/average. Note that the calibration i s different from that i n the ea r l i e r figures. 69. 100 m s e c Fig. 14 Averaged emitted potentials for one subject (MP) at multiple recording sites. Responses to omitted squares and to omitted els are shown superimposed (n=30 trials/averaged waveform). TABLE V T I I : Amplitude and latency (mean and standard deviation) of for responses to omitted and present stimuli of paradigm 2. A l l measurements made on averages of 10 t r i a l s . Number of averages measured for each subject indicated i n parenthesis. Omitted Stimuli: subject latency (msecs) amplitude (uv) MD (9) 330 - 35 1 2 - 2 SP (9) 335' - 75 13 - 2 MP (14) 360 - 45 17 - 5 ML (10) 335 - 30 10 - 4 SC (10) 325 - 35 1 2 - 3 5 subject mean 337 13 Present Stimuli: MD 325 - 40 12 - 2 SP 305 - 35 12 - 4 MP 300 - 40 24 - 5 ML 330 - 25 16 - 3 SC 355 - 20 18 - 5 5 subject 16 TABLE IX: Amplitude and latency (mean and standard deviation) of N~ for + responses to omitted and present stimuli of paradigm 2. n = 10 averages for each subject, experimental data as Table VTI. Measurements made on the same Omitted Stimuli: subject MD SP MP ML SC latency (msecs) 95 - 15 120 - 40 130 - 25 110 - 25 9 0 - 3 5 amplitude ( v) 5 ± 1 6 - 2 1 0 - 5 6 ± 2 7 ± 2 5 subject mean 110 Present Stimuli: MD SP MP ML SC 135 - 20 150 - 10 125 - 5 125 - 20 140 - 10 5 ± 2 8 - 3 7 ± 2 6 ± 2 9 ± 3 5 subject mean 135 + The Student t test for differences between the mean latencies for each of the 5 subjects shows that the latency of the N 2 component for the present stimuli i s significantly longer (p = .05) than for the omitted stimuli for 3 of the 5 subjects (MD, SP, & SC). (t = 5.06, 2.30, 4.34 respectively which i s greater than t 1 8 ^  > Q 5 = 2.1) 72. It can be seen by inspection of the magnitude of the standard deviations given i n Tables VIII and IX that the mean amplitude of N 2 and P 3 at C z were not significantly different for the evoked and emitted potentials. These measurements were not done for the parietal and o c c i p i t a l data for the P., component because i t was of maximum amplitude at C for both present and emitted stimuli. They were not done for the N 2 component because i t was very obvious i n the evoked responses (see Figs. 7 & 9 for examples) and small or absent i n the emitted ones (as shown i n Fig. 14) at these locations. I t i s also obvious from inspection of Table VIII that the latency of the P^ peak i s not significantly different for the evoked and emitted potentials. However the N 2 peak i s significantly e a r l i e r (Student t test p = .05) i n 3 of the 5 subjects. I t was also noted that i n the individual experiment averages, the amplitude of P^ did not correlate well with the subject's performance indicated by the number of button presses made within the appropriate time period. 5.3.2 Averaged "shifted" emitted responses. The "shifted" averages prepared as described i n the Methods chapter are shown i n Figs. 15 & 16. In Fig. 15, the waveform recorded at C i n each of the five subjects as i t appears after the shifting i s shown and Fig. 16 ill u s t r a t e s the "shifted" responses of two of the subjects at five recording sites. In both these figures an averaged response for omitted squares i s shown superimposed on one for emitted e l s . Since the shifting procedure was performed on f i n i t e 512 msec, samples, the average prepared 73. 2 0 JJV + ' 1 0 0 msec Fig. 15 Shifted emitted potentials for each of the 5 subjects recorded from C . Responses to omitted squares and to omitted els are shown superimposed. The arrow indicates the point to which the P 7 peaks were moved prior to averaging. 74. 20 uv + 100 msec 0 0 , 0 i Fig. 1 6 Shifted emitted potentials for two subjects (MD & SP) at multiple recording site s . Responses to omitted squares and to omitted els are shown superimposed. n=30 trials/average The arrow indicates the point to which the P^ peaks were moved prior to averaging. after the shifting i s only v a l i d for the middle time segment i n which a l l of the t r i a l s overlapped. Approximately 100 msec, was lost at each end because some of the t r i a l s were shifted as much as 100 msec, forward or backward along the time axis. The point to which a l l the peaks were moved i s marked i n the "shifted" averages of Fig. 15 and 16. I t occurs at 340 msec, after the time of the expected occurrence of the stimulus. The mean s h i f t for a l l of the waveforms was -10 msec, and the standard deviation of this mean s h i f t was 60 msec. This same shifting procedure was applied to the responses e l i c i t e d by the square and e l when they did appear (in paradigm 2 for eight experiments). The mean s h i f t for these waveforms was +40 msec, and the standard deviation of this mean s h i f t was 60 msec. This indicates that the variance of the P^ peaks i n the individual t r i a l s was the same for both the present and omitted stimuli, but that the peak tended to be ea r l i e r i n the responses to the present stimuli. The values reported i n Table VIII were determined by measuring the peaks i s unfiltered average waveforms for both the present and absent stimuli. I t i s evident from the Table that the latency for the P^ peak from the present stimuli i s e a r l i e r than the one to the omitted stimuli; but the difference i s much smaller than that reported above. The shifting technique using the peak of P^ as a marker did not make the ear l i e r negative components any clearer than they were i n the simple averages of the responses to both the present and omitted stimuli. I t can also be seen i n a l l of the data shown i n Figs. 13, 14, 15 and 16 that 76. there were no consistent differences between the responses to emitted squares and emitted els (as there were i n the responses to present squares and present e l s ) . 5.3.3 Control studies; motor potentials, EOG contributions and topography. In order to ascertain i f there was any contribution to the emitted response waveform from a motor potential resulting from the button pressing movement, a few experiments were run with each subject i n which no button press was required; the subjects were requested instead to note and count (silently) the number of omitted stimuli. The emitted responses shown i n both Figs. 17 and 18 are from experiments where the subject received these instructions. I t can be seen that the responses to the omitted stimuli recorded at C z are similar to the ones shown for the same subjects i n Fig. 13, and the findings were similar i n other experiments run without the button press. I t can thus be concluded that the motor response did not affect the emitted potential waveform. The EOG was also recorded i n a few experiments on one of the data channels to check for any contribution to the vertex a c t i v i t y by eye movement that may not have been excluded by the artefact rejection sub-routine of the program. I t was found that the EOG appeared f l a t i n the presence of a well defined component at the vertex. Fig. 17 i l l u s t r a t e s one of the experiments run with the EOG control. 77.*' • 10 JJV + 100 msec Fig. 17 Responses to omitted stiinuli with EOG recorded. Subject instructed not to make a button press. Arrow indicates time of expected occurrence of the stimulus. n=20 t r i a l s / average. (Subject MP, Exp. 71) 78. 10 ^ I 100 msec Fig. 18 Responses to omitted stimuli recorded in frontal regions. Subject instructed not to make a button press. Arrow indicates time of expected occurrence of the stimulus. n=40 trials/average. (Subject SP, Exp.55,56) Finally, i n a few experiments, recordings were made from frontal (F^, F^ and F^) electrodes to observe whether there was any response to the omitted stimuli at these locations). I t was found that the component was present at F but was smaller i n amplitude there than at C z and was not seen at F 3 or F^. Fig. 18 shows the emitted potentials for one of the subjects with recording at frontal sites compared to activ i t y at C and P . z z 5 . 3 . 4 Summary of the findings for omitted stimuli. (1) A P , component maximum i n amplitude at C was found i n the *j z emitted potentials that was of the same amplitude and latency as the P^ component at C z for present stimuli. This P^ component i n the emitted potential also appeared to have the same topographical distribution as the one for the evoked responses. (2) Control experiments showed that neither motor potentials or EOG could have been the source of the recorded waveforms. (3) A negative "N2" component was seen also maximum i n amplitude at C z, that had the same amplitude but was of ea r l i e r latency than the N 2 component of the response to present stimuli at the same location i n 3 of the 5 subjects. This "N2" component was not seen consistently at parietal or o c c i p i t a l locations i n the emitted responses, whereas the component of the evoked responses was much larger there than at C„. 80. (4) Shifting the single emitted response t r i a l s on the time axis prior to averaging to make the peaks coincide did not enhance the ea r l i e r components. (5) No consistent differences were seen between the emitted potentials e l i c i t e d by the omitted square and those e l i c i t e d by the omitted e l . 81. Chapter IV DISCUSSION. The main purpose of this study was to demonstrate and quantify the differences i n the visual evoked responses to different geometric forms. Evoked response differences were shown by (1) measuring the amplitudes of selected components, (2) computation of a descriptive ratio called A by John et a l . (1967) i n a study similar to part of this one, and (3) the performance of c l a s s i f i c a t i o n functions computed by Stepwise Discrimnant Analysis on new data. The s p e c i f i c i t y (for stimulus shape) of the responses i n the interval when the subject expected but did not see one of the geometric forms was also studies i n recordings from o c c i p i t a l , parietal, central and vertex regions. Some "emitted" responses were found mainly i n central regions but these were not specific for the expected shape. 6.1 Studies of Evoked Responses and Stimulus Meaning. Although the responses to the four shapes were found to be different, these differences were not always as obvious as the il l u s t r a t i o n s i n reports of other workers (who have done similar studies of VER's e l i c i t e d by different shapes related to meaning, John et a l . , 1967; Austt et a l . , 1971; Buchsbaum & Fedio, 1969) have suggested they might be. Other authors (Lifshitz, 1966; Chapman & Bragdon, 1964) who also studied the effects of stimulus meaning used stimuli with such differing physical properties that their conclusions can be questioned. 82. In the study of John et a l . (1967), specific components of the responses to the square, c i r c l e and diamond (analagous to the stimuli used i n the present study) were not considered. From the il l u s t r a t i o n s provided i t appeared to be the act i v i t y frcm 120 to 250 msec, that was different for the square and diamond. This i s approximately the same latency range as the differences reported for the shapes i n the present study. However no measures of a l l of the individual averaged responses were reported i n their paper. Only a ratio-type descriptor X was given, and even this s t a t i s t i c was reported only for selected subjects. In a second report from the same laboratory (Herrington & Scheidau, 1968) of a study i n which the subjects were instructed to think " c i r c l e " or "square" when a blank flash appeared; the response to the imagined shape was reported to be similar to that of the same real stimulus. I t was concluded by the authors i n both these studies that the visual evoked response waveshapes were determined by the specific cognitive processes underlying the recognition of the shape rather than the physical properties of the stimulus. Austt et a l . (1971) also reported that there were differences between visual evoked responses to different shapes. They even showed differences i n responses to the same figure when i t was perceived differently (a Necker cube). M l of the differences shown i n their figures were also i n the 150 to 250 msec, range. They recorded both vertex and oc c i p i t a l evoked responses and found that only the latt e r were shape specific. However their report did not provide any analysis of the reported differences, nor did they indicate that more than one subject was tested. 83. In another study of the effect of stimulus meaning on VER's, Buchsbaum & Fedio (1969) found different response configurations for dot patterns making up words or designs or random patterns. They attempted to quantify their results with a ratio-type descriptor ' V (similar to the X descriptor of John et a l . (1967)) computed for s e r i a l time segments of the response. The differences i n the waveforms that they found were mainly later than 300 msec. Therefore i n two of these three published studies on the effect of the meaning of the stimulus on the VER, the differences shown have been i n the same latency range (110 to 300 msec.) as were the differences found i n this research. In the third (Buchsbaum & Fedio, 1969), the major differences were demonstrated i n later components but their stimuli were dot patterns rather than simple geometric shapes and they were comparing mainly verbal and non-verbal material. In spite of the importance of these authors' conclusions with respect to concepts of the physiological substrates of cognition the analysis techniques and experimental controls i n these works were very superficial. When more detailed consideration i s given to the analysis and stimulus variables as i n the present study, the evidence does not appear to be as clear as these authors suggest. 6.2 Techniques for the Analysis of Evoked Responses. Quantification for s t a t i s t i c a l purposes requires many replications of the observations i n order to obtain an estimate of variance (which i s essential in establishing whether differences between two sets of observations are 84. significant or could have occurred by chance). Consider the observation that different shapes may evoke different responses. If the differences do refl e c t "central processes" related to meaning, when the stimuli must be presented a great number of times, the differences i n central processes may disappear. Both John et a l . (1967) and Austt et a l . (1971) used averages of only 25 to 50 t r i a l s and reported that i n averages of more t r i a l s the differences i n the responses were no longer seen. They suggested this was due to habituation i n the subject. An alternative to doing many test replications with one subject to obtain adequate data for estimating v a r i a b i l i t y i s to run a small number of t r i a l s with many subjects. However, this adds the well recognized problem of inter-subject v a r i a b i l i t y i n response configuration (Regan, 1972) to the problem of i n t e r t r i a l v a r i a b i l i t y . In this study, both approaches were used and compared to establish whether there were significant differences between responses to different shapes. A large number of subjects (13) were tested one time each (30 t r i a l s of each shape) to assess visually the configurations of the responses and to measure the latency and amplitude of the obvious peaks. Then a small number of subjects were tested repeatedly i n short experiments at well spaced intervals to gather enough t r i a l s for seme alternative s t a t i s t i c a l analysis techniques. The long intervals (weeks to months) between runs were presumed to minimize boredom with the stimuli. Consequently the results reported here based on visual analysis include a consideration of inter-subject v a r i a b i l i t y , and those based on the X ratio and Stepwise Discriminant Analysis techniques are affected more by v a r i a b i l i t y between replications for each subject. 85. 6.2.1 Visual Analysis. On visual examination of the data (see Fig. 7 for an example) the differences between the responses to the shapes appeared to l i e between 120 and 300 msec, latency. When the three principal peaks of the response were measured for a l l subjects and shapes, the amplitudes of the two components at 148 and 219 msec, were (statistically) significantly different only when comparing the responses of the square to the el' s , or the responses of the c i r c l e to the omega's. By these amplitude c r i t e r i a , neither the square and c i r c l e , nor the omega and e l were different from each other. The possible explanation for these observations w i l l be discussed after results from the other analysis techniques are considered. 6.2.2 X Values The mean X values reported i n Table VII were a l l larger than 1.0 which indicates that the averages for the different shapes compared were on the whole more different than the replications of the averages for the same shapes. However, the distribution of the values (see Fig. 12) was quite different from that of John et a l . (1967) who reported that 90% of their values f e l l between 1.0 and 1.7. In this study only 23% f e l l i n this range, with 37% of the values larger than 1.7 and 40% smaller than 1.0. On examination of the responses corresponding to individual X values i t was found that when two averages differed as much as those i n the il l u s t r a t i o n s of John et a l . (1967), and as long as the replications of the same shape were similar (the denominator of the r a t i o ) , X values of 2 or more resulted. 86. This suggests that i n much of their data the differences were not as easily v i s i b l e as those i n their i l l u s t r a t i o n s . A possible explanation for the difference between the X values reported here and those of John et a l . (1967) may be related to their use of a larger number of t r i a l s (30) to compute their averaged waveforms. A lower 'n' would make the averages "noisier" and decrease r e p l i c a b i l i t y . This would tend to make both the numerator and denominator of the X s t a t i s t i c larger and more variable, and thus explain why the X values i n this study showed such a scattered distribution. This ratio-type of s t a t i s t i c provides a quantitative estimate of the differences i n the data and does give a way of summarizing findings without having to i l l u s t r a t e a l l of the responses recorded. However, since i t s s t a t i s t i c a l distribution i s not known, i t i s not possible to decide i f the resulting values are significantly greater than 1.0, the value expected i f there was no real difference between the responses being compared. Also i t does not give any indication of where i n the response the differences occur unless i t i s computed for s e r i a l small time segments as Buchsbaum & Fedio (1969) do for their £ values. Even then this rather tedious technique only locates which of the a r b i t r a r i l y chosen intervals i s most different. 6.2.3 Stepwise Discriminant Analysis. Stepwise Discriminant Analysis (SWDA) i s a much more powerful and complex quantitative technique that does determine where the differences i n the responses l i e , and i s based on theoretical considerations which allow some 87. estimate of whether differences between two groups are significant provided the assumptions of the theory are met. (See Appendix B for details and an explanation of SWDA). I t uses single t r i a l data rather than averages and thus can measure the variance of the single t r i a l s included i n the averages, which i s ignored i n other techniques, and which i s important i n determining the s t a t i s t i c a l significance of the findings. In discussions of possible quantitative methods for analysis of evoked potential differences, Donchin (1966, 1969a, 1969b) suggests that the most useful one i s the Stepwise Discrirtiinant Analysis Technique provided i n the UCLA BMD:07M computer program. He gives an extensive explanation of the theoretical basis for this application as well as a c r i t i c a l consideration of alternate analysis techniques. The BMD:07M program calculate some s t a t i s t i c s (e.g. a U value) which can theoretically be used to determine whether the separation of the two groups of data for which i t has determined discriminant functions i s s t a t i s t i c a l l y significant. However Lachin & Schacter (1974) suggest that the BMD:07M exaggerates differences between groups and therefore that i t s results must be interpreted cautiously. In the applications he has so far reported Donchin (Donchin & Cohen, 1967; Donchin, Callaway & Jones, 1970; Donchin & Herning, 1975) has shown and tested i t s use mainly as a quantitative method of determLning where the differences occur i n the responses (Donchin & Cohen, 1967) and as a quantitative measure of the presence or absence of a response in threshold determination studies (Donchin & Herning, 1975) . In a study of the responses in schizophrenics and normals (Donchin et a l . , 1970) 88. i t was used to show there were no s t a t i s t i c a l l y significant differences between the groups. Gardiner & Walter (1969) also used i t i n evaluating the differences i n auditory responses under two stimulation conditions. SWDA as provided by the BMD:07M program was used to find and evaluate the differences between the evoked responses to the different shapes i n the present study. Because of the relative newness of this application of the technique, as summarized i n the previous paragraph, the results from BMD:07M were extensively evaluated and tested on new data. When the functions determined for each subject to discriminate between the pairs of e l and square or omega and c i r c l e were evaluated for new t r i a l s , they c l a s s i f i e d from 63% to 68% of them correctly (see Table VT) which i s significantly different from the proportion of approximately 50% which would be expected i f the cl a s s i f i c a t i o n were only being made on a random basis, (i.e. i f the functions were unable to separate the responses from the different shapes). Functions computed to discriminate between a l l four shapes c l a s s i f i e d from 38% to 40% of the single t r i a l s correctly which i s also significantly better than the 25% level expected by chance. It was therefore concluded that some of the stimulus shapes used did e l i c i t EEG responses that were suff i c i e n t l y d i s t i n c t to be separately identified by this technique. I t may be assumed that the evoked potential configurations were reflecting differences i n the underlying c e l l u l a r a c t i v i t y provoked by these stimuli. The f a i r l y high incidence of misclassifications i s believed to be due to the non-event related EEG ac t i v i t y present i n each single t r i a l submitted for c l a s s i f i c a t i o n , a c t i v i t y which i s largely cancelled out i n averaged potentials. 89. The technique o f t e s t i n g the SWDA f u n c t i o n s on new d a t a (that was n o t used t o c r e a t e the function) a v o i d s r e l y i n g on t h e s t a t i s t i c s such as the U v a l u e produced by the program t o measure the v a l i d i t y o f the i n t e r g r o u p d i f f e r e n c e s , and hence reduces the importance o f s a t i s f y i n g the t h e o r e t i c a l assumptions o f l i n e a r independence o f the v a r i a b l e s and G a u s s i a n i t y o f the i n t e r o b s e r v a t i o n v a r i a n c e . (See Appendix B f o r more d e t a i l s o f these a s s u m p t i o n s ) . The l a t e n c i e s o f t h e p o i n t s chosen by SWDA d i d c o r r e s p o n d approximately t o the same range o f l a t e n c i e s o f the components where the d i f f e r e n c e s were seen on v i s u a l i n s p e c t i o n and (as o t h e r authors have a l s o r e p o r t e d , D o n c h i n e t a l . , 1970) t h e r e was some v a r i a b i l i t y i n the l o c a t i o n o f the d i f f e r e n c e s among the d i f f e r e n t s u b j e c t s . 6.3 T o p o g r a p h i c a l d i s t r i b u t i o n o f the evoked r e s p o n s e s . 6.3.1 L o c a t i o n o f shape e f f e c t s . Although a l l o f the responses were r e c o r d e d a t s i x o r seven s c a l p l o c a t i o n s , i t was always t h e o c c i p i t a l l o c a t i o n s t h a t showed the c l e a r e s t and most r e p l i c a b l e responses and the b e s t v a l u e s i n SWDA t e s t i n g . Few o t h e r authors have r e c o r d e d from more t h a n one s c a l p l o c a t i o n i n t h i s type o f s t u d y . A u s t t (1971) compared o c c i p i t a l and v e r t e x responses and found d i f f e r e n c e s o n l y i n the o c c i p i t a l o n e s . John e t a l . (1967) and Buchsbaum & F e d i o (1969) b o t h used o c c i p i t a l e l e c t r o d e placements as have a l l o f the s t u d i e s on the e f f e c t s o f p h y s i c a l parameters (such as checkerboard s i z e and o r i e n t a t i o n ) on the evoked p o t e n t i a l . L i n d s l e y , W i l s o n & S e a l e s (1974) r e p o r t e d changes i n a l a t e P 9 wave i n the responses t o v i s u a l s t i m u l i f o l l o w i n g the l e a r n i n g 90. of an association with the stimulus. These changes were largest over oc c i p i t a l regions, less marked but present over Wernicke's region and absent at the vertex. Begleiter et a l . (1973) used flash stimuli of different intensities which induced changes at C and not at 0 , but their paradigm included association z z of a flash with a tone, an auditory atimulus which may have affected the evoked response at C . In similar experiments using varying flash z intensities without tones, no comparison of 0 and C was even made; a l l z z results were reported only for the C recording s i t e . The topography of z the VER to flash i s quite different from that of the VER e l i c i t e d by a pattern (Leserve, 1973) and this may be the reason for the vertex showing changes that one might expect to see at the o c c i p i t a l electrodes with a pattern stimulus. Johnson and Chesney (1974) i n their study of the effect of meaning on the VER, found correlates of meaning only i n frontal and not i n o c c i p i t a l regions. They were using an ambiguous patterned visual stimulus so this i s rather surprising. The finding i s d i f f i c u l t to explain (the authors did not try) i n terms of known physiological-anatomical relationships. They used a small number of subjects, and i n most of them recorded only from o c c i p i t a l or frontal locations, so they may i n fact have been seeing only an effect peculiar to a few subjects rather than a general one of functional localization. 6.3.2 Distribution of response components. The topography of the response components in this study was not determined 91. in sufficient d e t a i l to allow precise localization of possible c o r t i c a l sources, but some comparisons can be made with other experiments i n which more detailed observations were made. Because of the complex anatomy of the primary visual cortex, even simple localization studies require stimulation of the hemi-fields of vision separately. The component (95 msec, mean latency) was largestin o c c i p i t a l and parietal regions and much smaller but s t i l l positive i n polarity i n central recordings. The latency and distribution of this cranponent correspond to "wave 1" of Lesevre's (1973) study and she also observed that this component did not change with different stimulus patterns. She suggested that this "wave 1" represents a mixed response of foveal c e l l s to luminance as well as to contrast; the experimental findings of this study would support the suggestion that i t i s a luminance response. The N 2 component (148 msecs. latency) which i n this study did show changes evoked by the different stimulus patterns was s l i g h t l y later than Lesevre's (1973) "wave 2" (which she also demonstrated to be a pattern - specific component; the latency difference could be due to variations between stimulation techniques). The fact that the wave reversed i n polarity anteriorly i n some of the subjects, suggests that i t s source must be either a single dipole laying p a r a l l e l to the recording surface between the o c c i p i t a l and central regions, or a combination of two or more dipole sources oriented such that the relative contribution of each one changes as the recording site i s moved anteriorly. This lat t e r p o s s i b i l i t y i s the one suggested by Jeffreys & Axford (1972b) for their C II component mapped for upper and lower h a l f - f i e l d patterned visual stimulation. 92. The P 2 component (219 msecs. latency) occurred with the same polarity at a l l recording locations but was more variable in latency at the different sites than the earlier components. This component appears to be comparable to Lesevre's (1973) "wave 3", considered by her and others (Sphelmann, 1965) to be part of the pattern-specific VER. In the present study this component also showed some amplitude variation according to the stimulus pattern used. The origin of this peak has not yet been explained by any authors. Its wide distribution over the scalp and its variability suggest that i t reflects more widespread involvement of cortical generators than the previous two components. There were no significant hemispheric asymmetries in evoked response latency or amplitude nor any indicated by the performance of the discriminant functions. The lack of asymmetry i s consistent with the findings of Harmony, Ricardo, Otero, Fernandez, Llorente & Valdes (1973) for flash stimuli and of Lesevre (1973) and Jeffreys & Axford (1971a & b) for patterned visual stimuli. It is in contrast to studies by other authors (Davis, 1975; Wood, Goff & Day, 1971; Buchsbaum & Fedio, 1969) which have suggested that asymmetries in evoked responses might represent differential engagement of the hemispheres in cognitive tasks. However, the studies of Wood et a l . (1971) and Buchsbaum & Fedio (1969) both utilized linguistic stimuli which this study did not. The hemispheric asymmetry reported by Davis(1975) was in a measure of waveform coherence between different recording sites over each hemisphere, a parameter that was not considered in the present study. 93. 6.4 Intershape differences related to physical properties of the stimuli. For an approach to the question of what underlying mechanisms might account for the observed intershape differences, and for comparison with the findings of other workers using patterned visual stimulation, i t i s important to consider which of the four shapes' responses differed the most from each other. I t was pointed out above that by the c r i t e r i a of amplitude measurements, only the square versus e l and c i r c l e versus omega were significantly different; the square and c i r c l e or e l and omega were not. The X s t a t i s t i c s were only computed for the two 'different' pairs. In the results from SWDA, when a l l four shapes were included for the discriminant function computations, the best separation was again between the familiar (square and circle) and unfamiliar (el and omega) shapes. It can be seen i n Table IV that i t was the c i r c l e and square which were most frequently confused with each other (15% of the classifications) and the omega and e l (12.5% of the classifications) were next. As noted i n the Results chapter for one subject (MP) a function to separate the square and c i r c l e c l a s s i f i e d 59% of 290 new t r i a l s correctly, a much lower percentage than the functions for other pairs correctly c l a s s i f i e d ; but i t i s s t i l l significantly better than the 50% that would have resulted from a random cla s s i f i c a t i o n . Some of the other possible pairs of shapes were run i n the BMD:07M for some of the subjects, and the U values were always higher than for the square-el or circle-omega discriminations indicating a poorer separation. Hence i t can be concluded that there were real and significant differences between the oc c i p i t a l evoked response configurations of the familiar shapes 94. (square and circle) and the unfamiliar ones (el and omega). These differences were shewn by measuring the N 2 and P 2 components, by ccmputing X ratios and by SWDA. There also appeared to be some small differences between the square and c i r c l e i n one subject (MP) by the discriminant analysis technique as shown in Table V, but because these did not appear to be consistently present i n the other subjects, a complete quantitative analysis of them was not done. In assessing what variables might contribute to the findings summarized i n the previous paragraph, the evidence of other studies using patterned visual stimulation can be considered. The presentation of a l l four shapes i n random order i n a single run allows general correlates of the subject's state such as arousal and attention, which are known to affect the response configuration, to be eliminated. The projector-to-T.V. monitor system insured that the contrast lev e l , focus, and size of the stimuli were always equal for the four shapes, and they a l l had very nearly equal lengths of contrasting, borders and luminance on projection. Thus, although various authors have reported changes i n the evoked response configuration due to each one of these factors, (Rietveld et a l . , 1967; Regan, 1972) none of them could account for the differences demonstrated i n this study. Much of the work using different patterned stimuli has been done with checkerboards or gratings because with this type of pattern the number of corners or lines can be changed without varying the t o t a l luminance of the stimulus. Sphelmann (1965) showed that the "pattern type" response was evoked by a variety of patterns including v e r t i c a l and diagonal stripes as well as checkerboards. The only difference i n responses to the different stimulus patterns that he noted was i n a positive component at 180 to 250 msec, in the VER's e l i c i t e d by the pattern of checks. This component increased i n 95. amplitude as the number of squares i n the f i e l d increased. Rietveld et a l . (1967), i n a more detailed study of responses to checkerboard-type patterned stimuli, also found a negative-positive complex between 120 and 200 msec, which was of maximum amplitude with check sizes subtending 27' of visual angle and which decreased i n amplitude with larger checks. They also noted that increasing stimulus luminance increased the amplitude of the responses and that only the pattern i n the central 4° of the visual f i e l d was responsible for the pattern evoked response. Harter & White (1968) confirmed the existence of the negative-positive complex from 100 to 200 msec, i n the response evoked by checkerboard stimuli and also reported that i t s amplitude i s very sensitive to the size of the checks as well as to the sharpness of the focus. They also noted that the closed type of patterns (checkerboards) produced larger responses than open ones (stripes). Could these findings explain the differences demonstrated here between square and e l responses? The e l when compared to the square would have some of the contours of a smaller checkerboard. However i n this analogy, the square as a check would subtend a visual angle of approximately 3.6° and the e l one of 1.8°. The difference between the responses to these two sizes was very small i n the study of Rietveld et a l . (1967). Harter & White (1968) did not use checks any larger than 46'. In the present study, when the stimuli were shown at half size (hence reducing the visual angle of the square to 1.8°) no change i n the response configurations was evident. Some authors (Regan, 1972; MacKay, 1969) have suggested that the presence of corners i n the stimulus pattern i s an important determinant of evoked response configuration, but definite evidence for this i s not yet available. 96. The demonstration of Rietveld et a l . (1967) that checkerboards produce larger responses than stripes and that blurring the corners of the checker-board changed the evoked responses are cited as support for the suggestion of Regan and MacKay, but both of Rietveld's findings could be explained on the basis of changes i n to t a l contrasting border length. Furthermore, the present study has shown that evoked responses to c i r c l e and square were quite similar, although the shapes differed by four corners, whereas the omega and c i r c l e differed by only two corners and produced quite different responses. The presence i n the visual cortex of orientation detectors that affect the evoked response configuration has been suggested by studies both of steady state (Campbell & Maffei, 1970) and more recently of transient evoked potentials (Yoshida et a l . , 1975). These authors were using gratings or striped stimuli and their results suggest that the findings of John et a l . (1967), of a difference between responses evoked by squares and diamonds could have been due merely to the change i n orientation of the stimuli. However other groups (Kakigi et a l . , 1972; Rietveld et a l . , 1967) have reported no change at a l l i n the response after rotating a square stimulus through 45°, so the question of the existence of orientation detectors whose ac t i v i t y i s reflected i n the c o r t i c a l evoked response i s s t i l l unresolved. Harter's (1970) findings on the effect of re t i n a l eccentricity and check size however may be more relevant and could at least p a r t i a l l y explain why complex patterns may evoke different responses. He showed that with a 90' check there was about a 3yv drop i n amplitude of the 180 msec, positive component when the area stimulated was changed from the central 0 to 1.5° 97. to the more peripheral 1.5° to 4.0° of the visual f i e l d . In the present experiments both the e l and omega had more contrasting borders i n the central 1.5 of the subject's visual f i e l d than the c i r c l e or square did. A difference was found between the responses to these two pairs of shapes i n a positive component around 220 msec, latency. The difference between the latency of these components and Harter's could be due to variations i n stimulus luminance and pattern structure. The spatial frequencies of geometric forms may also be important determinants of the evoked response configuration. Blakemore & Campbell (1969) and Campbell (1974) suggested that i n the visual system there are channels tuned to spatial frequency and that patterns may be transmitted through these channels, broken down into spatial frequency components, analgous to Helmholz's (1877) hypothesis of transmission of sounds broken down into temporal frequency components i n the auditory system. Campbell's laboratory has provided some evidence that steady-state evoked potentials r e f l e c t the spatial frequency of the visual stimulus (Blakemore & Campbell, 1969; Campbell & Robson, 1968) and a recent paper by Musso & Harter (1975) recording transient evoked potentials confirms the findings. These latter authors found a masking effect on the 110 msec, negative component of the evoked response to the second stimulus of a pair only i f the f i r s t stimulus contained the same check sizes as the second. The procedure to obtain a precise, numerical estimate of the spatial frequencies of the shapes used i n this study i s very complex and i s considered to be beyond the scope of this discussion. However, even on visual inspection i t i s clear that there are more spatial frequencies in the e l shape than i n 98. the square (because of the smaller distance between the vertical lines and the horizontal lines of the el shape compared to the square). The same would be true for the omega compared to the circle. The comers contained in the shapes tend also to increase the number of spatial frequencies in the figure, but their relative contribution is uncertain. Thus i t may be that the greater number of spatial frequencies contained in the el and omega compared to the square and circle is an important determinant of the higher amplitude of some of the components of the responses to the unfamiliar shapes that was seen in this study. In summary, there i s evidence from other studies that at least some of the differences found between the responses to the shapes could be due to the physical properties of the stimuli. The difference in amount of contrasting border present in the central 1.5° of the subject's visual field and/or the difference in spatial frequencies between the square or circle and the e l or omega are the most likely c r i t i c a l physical properties for these shapes. However, a l l of the effects so far reported of physical properties of the stimulus on the evoked response occur between 100 and 220 msec, latency. Differences in the responses to the four shapes used in this study were found by SWDA up to 270 msec, after the stimulus presentation., While i t might be argued that differences in luminance and size of these shapes, as compared to those used in other studies, could account for this finding, an alternative explanation must also be considered; the components after 220 msec, were reflecting the meaning rather than the physical properties of the stimulus. 99. 6.5 Effects of stimulus meaning. 6.5.1 Present stimuli. In comparing responses to present stimuli from paradigm 1 and 2, i t was found that the imposition of a task changed the configuration of the evoked potential only after 220 msec, latency. A division of the response into earlier, stimulus-related components and later task-related components i s i n keeping with the evidence of numerous investigators' reports of the properties of P^ or "late positive components" (Ritter & Vaughan, 1969; Hillyard, 1974). These later components have been shown to be sensitive to a wide variety of task-related variables (Beck, 1975; Squires et a l . , 1975b and section 2.3 of this dissertation). Very l i t t l e experimental evidence has been provided to date that clearly separates the effects on the evoked response of non-contextual stimulus meaning from physical properties of the stimulus and from i t s task-related meaning. This f i r s t separation i s perhaps only possible for stimuli that can be verbally defined. In studying information processing, i t i s usually assumed that between the external stimulus and behavioral response, a number of component operations take place at different stages, either simultaneously or i n p a r a l l e l . Chapman (1974) described a study of evoked responses using factor analysis during information processing tasks. The purpose was to find specific components of the evoked response associated with certain aspects of information processing. He reports i n his preliininary findingsseven components, selected by factor analysis, each of which could be associated with different cognitive processes involved in his complex paradigms. 100. Johnson & Chesney's (1974) use of an ambiguous visual stimulus that could be interpreted as a number or letter was a good design to isolate the meaning component related to verbal definition, but their evoked response results were inconclusive because of the small number of subjects used. An analysis of the perceptual-cognitive and response elements of the task in the paradigms used i n the present study suggests the following sequence: the subject perceived the physical properties of the stimulus, then attached a l i n g u i s t i c label to i t , and then performed the task-related association i f there was one (i.e. counting or a motor response). For patterned visual stimuli, i n this and the other studies reviewed above, the o c c i p i t a l evoked potential has been definitely shown to have components related to the f i r s t and third elements of this information sequence. The later, task-related components may also be clearly seen at the vertex as demonstrated both i n this and some other studies, suggesting that associative processes involve widespread c o r t i c a l areas. The intermediate step of attaching stimulus (linguistic) meaning, was not su f f i c i e n t l y isolated i n these experiments to allow any conclusions about i t s evoked potential correlates. However, because of their unfamiliarity this intermediate step presumably involved more complex processing for the e l and omega than for the more familiar square and c i r c l e , and this could have been an important determinant i n the differences i n the waveforms between 220 and 270 msec, i n the no-task condition. There have not been any reports i n the literature of evoked potential studies that provide any more definitive answers to the question of whether this particular step i n the sequence i s represented i n the c o r t i c a l evoked response. 101. 6.5.2 Omitted stimuli. I t was shown in this study that "emitted potentials" could be recorded to missing visual stimuli. The use of this sensory modality for obtaining emitted potentials has been reported by very few investigators to this date (Ruchkin & Sutton, 1973; Weinberg et a l . , 1969; Ritter, Simson & Vaughan, 1974). In agreement with Ruchkin & Sutton (1973) a positive component was recorded about 330 msec, after the time of the expected but emitted stimulus with maximum amplitude at C . This wave was somewhat ea r l i e r i n latency at 330 msec, i n this study compared to 400 msec, i n their report. Also i n contrast to their results, the amplitude of the 'P^' component was not found to be of significantly different amplitude under the real stimulus and omitted stimulus conditions. In their review of the phenomena of emitted potentials, Sutton et a l . (1974) called this positive component the P 3 component because of the evidence that i t represents the same processes as the P^ component of evoked responses. (See section 2.3 of the Literature Review chapter for further discussion). Weinberg et a l . (1970) used frontal intracerebral and subdural electrodes so their waveforms are not really comparable to those of the present study. They found a slow positive component, that occurred at shorter latency i n the emitted responses than i n the evoked ones, a finding not repeated i n this study. Ritter et a l . (1974) also reported a P^ component of the emitted response similar to the one seen i n this study. They described i n addition a negative component preceding the P^ ac t i v i t y which had a different distribution for 102. auditory and visual omitted stimuli. No other investigators studying the cortical responses to omitted stimuli in any modality have as yet reported the presence of negative components preceding the activity. In the emitted responses of the present study, there appeared to be a negative component (called N 2 because i t had a similar latency and appearance to the N 2 component in the evoked responses) preceding the positive component, but i t was not always possible to identify i t clearly. In the single experiment averages (10 trials) a negative component could be identified in a l l cases in central channels only, and sometimes even here i t was of such small amplitude that i t was difficult to separate from the background noise. The mean and standard deviation of the latency of this peak compared to the mean and standard deviation of the N 2 component in the evoked responses given in Table IX. The variance of the component in the emitted potentials is much higher than in the evoked ones. In the evoked responses the N 2 component was prominent at other recording locations (parietal and occipital), but was not seen at a l l in the emitted responses at these other sites. Two possibilities must be considered; either this negative component was not a part of the emitted responses but resulted from spurious non-event related activity; or i t does represent some brain processes related to the internal representation of the stimulus. An assessment of the f i r s t possibility requires reconsideration of the methods of defining an evoked response component as "separate" from the background non-event related EEG. The traditional method of distinguishing event-related activity from "noise" by averaging requires the event-related 103. activity to be time-locked to the event. In work with omitted stimuli this requirement cannot be met, but an alternative method for signal-to-noise reduction has not been devised. In this study i t was hypothesized that a l l a c t i v i t y related to the experience of the omitted stimulus might have a fixed time relationship to the peak (which could be identified i n each single t r i a l sample). Consequently a post hoc time-locking technique was devised. Each single t r i a l sample was shifted on the time axis so that the P^ peaks coincided prior to averaging, but the earl i e r negative components did not become any more obvious i n these "shifted" averages. For control purposes the shifting procedure was t r i e d on samples following real stimuli as well. The P^ peaks were found to be very nearly as variable in time as i n the emitted stimulus responses. It was concluded that the latency of the P^ peak i t s e l f i s not constant even i n the responses to a real event. Thus the fact that the elusive negative ac t i v i t y of the emitted response was not time-locked to i t i s not surprising. The question of whether the emitted potential consists of anything more than a P^ component must be considered to be as yet unresolved. The paradigms used by the different investigators have varied. Perhaps they are the important determinants of the presence of early negative a c t i v i t y . Alternatively, new techniques for separating signals from noise may be needed to firmly establish the existence of this component. If the second p o s s i b i l i t y suggested above could be established to be the correct one, the data of this study suggest that the processes this component reflects i n occ i p i t a l and parietal regions are related to sensory input and 104. in central regions are related to the conceptual representation of the geometric shapes. Moreover, i t appears that whatever brain processes i t represents, this central ac t i v i t y i s not specific for the particular shape that i s emitted. This i s i n contrast to the shape specific a c t i v i t y that appears i n the presence of a pattern stimulus i n the parietal and oc c i p i t a l regions. Chapter VTI SUMMARY AND CONCLUSIONS The recording of cerebral evoked potentials has been widely used as a method of studying correlates of sensory processes i n human subjects. In the visual system i t has been established that certain physical parameters of the stimulus including intensity, focus, and size of the elements of a pattern have significant effects on the configuration of the evoked response as recorded from the o c c i p i t a l region. While claims have been made i n a few publications, to this date the question of whether any part of the evoked response i s affected by higher perceptual processes rather than ones more directly related to stimulus input has not yet been resolved. In this study, the evoked potentials were recorded from multiple scalp, locations using four different geometric stimuli (a square, c i r c l e , e l and omega shape) under two experimental conditions; i n the f i r s t the shapes were presented to the subject at random intervals and i n random order with no meaning assigned to them, and i n the second the shapes were presented i n a fixed rhythmic sequence and were important cues for a task assigned to the subject. During this second experimental paradigm, two of the shapes occasionally f a i l e d to appear i n their usual time interval i n the sequence and the subject was required to signal such omissions by a button press. Cerebral e l e c t r i c a l a c t i v i t y during the time interval of these omitted but expected stimuli was recorded and averaged and these responses (called emitted potentials) were also analyzed. There were significant differences between the evoked responses (in the occi p i t a l region) to the square and e l shapes and between those to the 106. c i r c l e and omega shapes. These differences were demonstrated by three measurement techniques: performance of the discriminant functions computed by SWDA i n classifying single t r i a l responses, the X descriptor of John et a l . (1967), and amplitude differences i n the ^ and components. The differences between the responses to the e l and omega and between those to the c i r c l e and square were less consistent. In parietal regions the responses were characteristic for the different shapes only i n some subjects, and i n central regions they were much more variable and definitely did not appear to be shape-specific. The task added in paradigm 2 changed the evoked responses e l i c i t e d by the square and e l shapes (the ones occasionally omitted). Under these conditions, the responses included a late positive component that was not presented i n the evoked responses recorded i n paradigm 1. This peak was of maximum amplitude at C and was small i n o c c i p i t a l regions. The emitted potential which was recorded i n the interval i n which the subject expected but did not see a stimulus, contained a small and variable early negative component and an obvious late positive ccmponent similar to the one seen i n the responses to the present square and e l i n this paradigm. Manipulation of the waveforms from the single t r i a l s prior to averaging showed that this negative component was not time-locked to the peak. The evoked potential differences found were believed to be due to two classes of variables; the physical characteristics of the stimulus (the contours i n the central 1.5° of the visual field) and task-related changes in the meaning of the stimulus. These two types of variables were responsible for changes i n the e a r l i e r and later (after 220 msec.) parts of the waveform 107. respectively. The existence of task-related changes in the evoked response has been established by other investigators, but this is the f i r s t study to clearly demonstrate and quantify differences in the earlier parts of the visual evoked response elicited by stimuli of this type. This is the f i r s t study to explore a variety of methods of measuring the differences between evoked potential waveforms, including a method that utilizes single trials. This technique, based on multivariate statistical theory is called Discriminant Analysis (as provided by the UCLA BMD:07M computer program run on the university IBM 370 computer). Its application to this type of problem has been suggested and described in a few other reports, but this is the f i r s t report of the performance of the method being empirically tested on this type of experimental data. Some programs were developed for a PDP 11/20 computer that controlled the presentation of the stimuli and the collection, storage and subsequent manipulation for analysis of the digitized evoked potential waveforms. These programs included routines for moving the responses on the time axis, so that the single trials could be individually altered prior to averaging. They could also be transferred to files on the university IBM 370 computer for more extensive or complex processing. The study is one of the few reports available describing the activity recorded in the interval when the subject is expecting but does not actually see a patterned visual stimulus. These emitted potentials were not found to be characteristic for the shape the subject was expecting, in contrast to the evoked potentials to the present stimuli. 108. On the basis of a c r i t i c a l consideration of the experimental findings and of reports of other investigations using patterned visual stimuli, a theoretical explanation for the underlying mechanism of these phenomena i s proposed. The evoked response reflects the following sequential processes occurring i n the cerebral cortex following the presentation of a patterned visual stimulus: The a c t i v i t y up to 100 msec, after the stimulus occurs i s a response to the luminance of the stimulus and i s not related to i t s pattern. This time period may be s l i g h t l y shorter or longer (by approximately 20 msec.) depending on the magnitude of the luminance of the stimulus. Between approximately 100 and 220 msec, the c o r t i c a l a c t i v i t y recorded reflects responses of c e l l s sensitive to contours i n the stimulus pattern, particularly to those i n the central (foveal) 1-2° of the visual f i e l d . Until this time point i n the response, this stimulus-related a c t i v i t y appears to be originating i n the o c c i p i t a l cortex. I t may be recorded at other locations, but the waveforms recorded there do not as f a i t h f u l l y r e f l e c t physical aspects of the stimulus. During the interval when the subject i s expecting but does not see the visual stimulus, a peak occurs between 100 and 220 msec, which can be recorded i n the vertex region. This vertex act i v i t y does not appear to be affected by the properties of the stimulus that was expected. While the definite origin of e l e c t r i c a l a c t i v i t y recorded from the vertex has not been established (Walter, 1964), i t i s suggested that these emitted potentials may refl e c t a deep c o r t i c a l or subcortical non-specific process related to some internal representation of the stimulus the subject was expecting. 109. After 220 msec, more extensive areas of cortex become involved i n the responses to both present and omitted stimuli, and i t i s at this stage of the process that the recorded waveform reflects task-related variables in the stimulus meaning. This late a c t i v i t y i s also of maximum amplitude at the vertex, but i t appears over parietal and frontal cortex as well. The evoked potential studies i n this experimental work provide direct evidence of a sequential temporal pattern i n the co r t i c a l events following the presentation of a complex visual stimulus. The data suggest that the early a c t i v i t y originates i n visual cortex and i s related to the luminance of the stimulus. The processes involving perception of pattern follow this e a r l i e r a c t i v i t y and also originate i n the oc c i p i t a l lobe. Finally come the events related to the recognition of contextual meaning, which involve much more extensive areas of cortex than the other levels of processing of visual information. BIBLIOGRAPHY ARDEN, G.B., & LEWIS, D.R.H. (1973) The pattern visual evoked responses in the assessment of visual acuitv. Trans. Ophthalamol. Soc. U.K. 93: 39-48. AUSTT, E. GARCIA, BUNO, W. Jr., & VANZULLI, A. (1971) Evoked potentials and central processing of visual information. Vision Res. Suppl. 3: 457-477. BECK, EDWARD C. (1975) Electrophysiology and behavior. Annual Rev. Psychology 26: 233-262. BEGLEITER, H., GROSS, M.M. & KISSIN, B. (1967) Evoked c o r t i c a l responses to affective visual stimuli. Psychophysiology 3: 336-344. BEGLEITER, H. & PLATZ, A. (1969) Evoked potentials: modifications by cl a s s i c a l conditioning. Science 166: 769-71. BEGLEITER, H., & PORJESZ, B. (1975) Evoked brain potentials as indicators of decision-making. Science 187: 754-755. BEGLEITER, H., PORJESZ, B., YERRE, C., & KISSIN, B. (1973) Evoked potential correlates of expected stimulus intensity. Science 179: 814-816. BISHOP, P.O., COOMBS, J.S., HENRY G.H. (1971) Interaction effects of visual contours on the discharge frequency of simple striate neurons. J. Physiol. 219: 659-687. BLAKEMORE, C. & CAMPBELL, F.W. (1969) On the existence of neurons i n the human visual system selectively sensitive to the orientation and size of retinal images. J. Physiol. 203: 237-260. BUCHSBAUM, M. & FEDIO, P. (1969) Visual information and evoked responses from the l e f t and right hemispheres. Electroenceph. c l i n . Neurophysiol. 26: 266-272. CAMPBELL, F.W. (1974) The transmission of spatial information through the visual system, i n The Neurosciences, Third Study Program, Ed. F. 0- Schmidt pp 95-104. MIT Press. CAMPBELL, F.W. & ROBSON, J.G. (1968) Application of Fourier analysis to the v i s i b i l i t y of gratings. J . Physiol. 197: 551-566. CAMPBELL, F.W., & KULIKOWSKI, J.J. (1972) The visual evoked potential as a function of contrast of a grating pattern. J. Physiol. 222: 345-356. CAMPBELL, F.W. & MAFFEI, L. (1970) Electrophysiological evidence for the existence of orientation and size detectors i n the human visual system. J. Physiol. 207: 635-652. CHAPMAN, R.M. (1974) Latent components of average evoked brain responses functionally related to information processing. Paper presented at the International Symposium on Cerebral Evoked Potentials in Man. CHAPMAN, R.M. & BRAGDON, H.R. (1964) Evoked responses to numerical and non-numerical stimuli while problem solving. Nature 203: 1155-57. CIGANEK, L. (1975) Visual evoked responses. Handbook of Electroenceph. clin. Neurophysiol. 8: 33-59. Ed. by A. Remond. CLYNES, M., & KOHN, M. (1967) Spatial visual evoked potentials as physiologic language elements for color and field structure. Electroenceph. clin. Neurophysiol. Suppl. 26: 82-96. COOPER, R., WINTER, A.L., CROW, H.J. & WALTER, W.G. (1965) Comparison of subcortical, cortical and scalp activity using chronically indwelling electrodes in man. Electroenceph. clin. Neurophysiol. 18: 217-228. CREUTZFELDT, O.D., & KUHNT, U. (1967) The visual evoked potential: physiological, developmental and clinical aspects. Electroenceph. clin. Neurophysiol. Suppl. 26: 29-41. CREUTZFELDT, O.D., WATANABE, S. & LUX, H.D. (1966) Relations between EEG phenomena and potentials of single cortical cells. I. Evoked responses after thalamic and epicortical stimulation. Electroenceph. clin. Neurophysiol. 20: 1-18. DAVTS, A.E. (1975) Spectral analysis of human evoked potentials. Thesis, University of B.C. DeVOE, R.G., RIPPS, H. & VAUGHAN, H.G. (1968) Cortical responses to stimulation of the human fovea. Vis. Res. 8: 135-147. DIXON, W.J. (1965) (Ed.) BMD - Biomedial Computer Programs. Univ. of California, Los Angeles. DONCHIN, E., (1969a) Data analysis techniques in average evoked potential research, in Average Evoked Potentials, (eds.) E. Donchin & D.B. Lindsley. NASA, SP-191, Washington, D.C. pp 199-236. DONCHIN, E..(1969b) Discriminant analysis in average evoked response studies: the study of single t r i a l data. Electroenceph. clin. Neurophysiol. 27: 311-314. DONCHIN, EMANUEL (1966) A multivariate approach to the analysis of average evoked potentials. IEEE Trans, on Biomedical Eng. BME-13: 131-139. DONCHIN, E., Ou^ LAWAY, E. & JONES, R.T. (1970) Auditory evoked potential variability in schizophrenia. II. the application of discriminant analysis. Electroenceph. clin. Neurophysiol. 29: 429-440. DONCHIN, E., & COHEN, L. (1967) Averaged evoked potentials and intramodality selective attention. Electroenceph. clin. Neurophysiol. 22: 325-35. 112. CONCHIN, E. & HERNING, R.I. (1975) A simulation study of the efficacy of stepwise discriminant analysis in the detection and comparison of event related potentials. Electroenceph. clin. Neurophysiol. 38: 51-68. ELUL, RAFAEL (1968) Brain waves: intracellular recording and statistical analysis help clarify their phsyiological significance, in Data Acquisition and Processing in Biology and Medicine. Proc. of the 1966 Rochester Conference, Ed. by K. Enslein. ESTEVEZ, 0. & SPEKREIJSE, H. (1974) Relationship between pattern appearance-disappearance and pattern reversal responses. Exp. Brain Res. 19: 233-238. FORD, J.M., ROTH, W.T., DIRKS, S.J., KOPELL, B.S. (1973) Evoked potential correlates of signal recognition between and within modalities. Science 181: 465-66. GARDINER, M.F. & WALTER, D.O. (1969) Differences between human evoked potentials elicited by the same acoustical stimuli during loudness discrimination tasks and pitch discrimination tasks, in Average Evoked Potentials. (eds.) E. Donchin & D. B. Lindsley. NASA, SP-191, Washington, D.C. GEISLER, CD. & GERSTEIN, G.L. (1961) The surface EEG in relation to its sources. Electroenceph. clin. Neurophysiol. 13: 927-934. HALLIDAY, A.M., & MICHAEL, W.F. (1970) Changes in pattern-evoked responses in man associated with the vertical and horizontal meridians of the visual field. J. Physiol. 208: 499-513. HARMONY, T., RICARDO, J., OTERO, G., FERNANDEZ, G., IlfiRENTE, S., VALDES, P. (1973) Symmetry of the visual evoked potential in normal subjects. Electroenceph. clin. Neurophysiol. 35: 237-240. HARTER, M.R. (1971) Visually evoked cortical responses to the on- and off-set of patterned light in humans. Vision Res. 11: 685-695. HARTER, M.R. (1970) Evoked cortical responses to checkerboard patterns; effect of check-size as a function of retinal eccentricity. Vision Res. 10: 1365-1376. HARTER, M.R. & SALMON, L. (1972) Intra-modality selective attention and evoked cortical potentials to randomly presented patterns. Electroenceph. clin. Neurophysiol. 32: 605-613. HARTER, M.R. & WHITE, CT. (1970) Evoked cortical responses to checkerboard patterns: effect of check-size as a function of visual acuity. Electroenceph. clin. Neurophysiol. 28: 48-54. HARTER, M.R., WHITE, CT. (1968) Effects of contour sharpness and check-size on visually evoked cortical potentials. Vision Res. 8: 701-711. 113. HEATH, R . G . & GALBRAITH, G . C . (1966) Sensory evoked responses r e c o r d e d s i m u l t a n e o u s l y from human c o r t e x and s c a l p . Nature 212: 1535-37. HELMHOLZ, H . (1877) ( r e p r i n t e d 1954) On the Sensat ions o f Tone. New Y o r k : Dover P u b l i c a t i o n s . HERRINGTON, R . N . & SCHNEIDAU, P. (1968) The e f f e c t o f imagery on the waveshape o f the v i s u a l evoked response. E x p e r i e n t i a 24: 1136-37. HILLYARD, S . A . (1974) L o n g - l a t e n c y evoked p o t e n t i a l s and s e l e c t i v e a t t e n t i o n . Paper p r e s e n t e d a t t h e I n t e r n a t i o n a l Symposium on C e r e b r a l Evoked P o t e n t i a l s i n Man. B r u s s e l s , Belgium. HOF, M.W. van (1960) The r e l a t i o n between t h e c o r t i c a l responses t o f l a s h and f l i c k e r i n man. A c t a P h y s i o l . Pharmacol. N e e r l . 9: 210-224. HUBEL, D . H . & WIESEL, T . N . (1968) Receptive f i e l d s and f u n c t i o n a l a r c h i t e c t u r e o f monkey s t r i a t e c o r t e x . J . P h y s i o l . 195: 215-243. HUBEL, D . H . & WIESEL, T . N . (1965) Receptive f i e l d s and f u n c t i o n a l a r c h i t e c t u r e i n two n o n - s t r i a t e v i s u a l areas (18 and 19) o f the c a t . J . N e u r o p h y s i o l . 28: 229-289. JASPER, H . (1958) The t e n twenty e l e c t r o d e system o f the I n t e r n a t i o n a l F e d e r a t i o n . E l e c t r o e n c e p h . c l i n . N e u r o p h y s i o l . 10: 371-375. JASPER, H . H . & STEFANIS, C . (1965) I n t r a c e l l u l a r o s c i l l a t o r y rhythms i n p y r a m i d a l t r a c t neurons i n the c a t . E l e c t r o e n c e p h . c l i n . N e u r o p h y s i o l . 18: 541-553. JEFFREYS, D . A . (1971) C o r t i c a l source l o c a t i o n s o f p a t t e r n - r e l a t e d v i s u a l evoked p o t e n t i a l s r e c o r d e d from the human s c a l p . Nature 229: 502-504. JEFFREYS, D . A . & AXFORD, J . G . (1972a) Source l o c a t i o n s o f p a t t e r n - s p e c i f i c components o f human v i s u a l evoked p o t e n t i a l s . I . Component o f s t r i a t e c o r t i c a l o r i g i n . Exp. B r a i n Res. 16: 1-21. J JEFFREYS, D . A . & AXFORD, J . G . (1972b) Source l o c a t i o n s o f p a t t e r n - s p e c i f i c components o f human v i s u a l evoked p o t e n t i a l s . I I . Component o f e x t r a -s t r i a t e c o r t i c a l o r i g i n . Exp. B r a i n Res. 16: 22-40. JENNESS, D. (1972) A u d i t o r y evoked-response d i f f e r e n t i a t i o n w i t h d i s c r i m i n a t i o n l e a r n i n g i n humans. J . o f Comparative and P h y s i o l . Psychology 80: 75-90. JOHN, E . R . (1974) Assessment o f a c u i t y , c o l o r v i s i o n and sharp p e r c e p t i o n by s t a t i s t i c a l e v a l u a t i o n o f evoked p o t e n t i a l s . Ann. Ophthalmol. 6: 55-66. JOHN, E . R . (1967) Mechanisms o f Memory. Academic P r e s s , New York pp 468. JOHN, E . R . & HERRINGTON, R . N . & SUTTON, S. (1967) E f f e c t s o f v i s u a l form on the evoked response. S c i e n c e 155: 1470-1471. 114. JOHN, E.R., SHIMOKOCHI, M., BARTIoETT, F. (1969) Neural readout frcm memory during generalization. Science 164: 1534-36. JOHNSTON, V.S., & CHESNEY, G.L. (1974) Electrophysiological correlates of meaning. Science 186: 944-46. KAKIGI, H., MIYAZAKI, M., & MORI, T. (1972) Effects of stimulus rotation, shape and visual angle upon human visual evoked response. Jap. Psychol. Res. 14: 153-157. KARLIN, L. (1970) Cognition, preparation and sensory-evoked potentials. Psychological Bulletin 73: 122-136. KILOH, L.G., McCOMAS, A.J., OSSELTON, J.W. (1972) Clinical Electro-encephalography. Butterworths (London). KINNEY, J.A.S., MCKAY, C.L., MENSCH, A.J., & LURIA, S.M. (1972) Techniques for analysing differences in VER's to colored and patterned stimuli. Vision Res. 12: 1733-1747. KLINKE, R., FRUHSTORFER, H. & FINKENZELLER, P. (1968) Evoked responses as a function of external and stored information. Electroenceph. clin. Neurophysiol. 25: 119-122. KULIKOWSKI, J.J. (1974) Human averaged occipital potentials evoked by pattern and movement. J. Physiol. 242: 70P-71P. LACHIN, J.M., & SCHACHTER, J. (1974) Methodology - on stepwise discriminant analyses applied to physiologic data. Psychophysiology 11: 703-709. LESEVRE, N. (1973) Potentiels evoques par des patterns chez 1'homme: influence de variables caracterisant le stimulus et sa position dans le champ visuel. Activites Evoquees et leur conditionnement chez 1'homme normal et en pathologie mental. Colloque Neurophysiologie Humaine. Paris INSERM pp 1-22. LI, C. & JASPER, H.H. (1953) Microelectrode studies of the electrical activity of the cerebral cortex in the cat. J. Physiol. 121: 117-140. LIFSHIFTZ, K. (1966) The averaged evoked cortical response to complex visual stimuli. Psychophysiology 3: 55-68. LINDSLEY, D.B., WILSON, G.F., SEALES, D.M. (1974) Evoked potentials and information processing. Electroenceph. clin. Neurophysiol. 36: 213. MacKAY, D.M. (1969) Evoked brain potentials as indicators of sensory information processing. Neurosciences Res. Prog. Bull. 7: No. 3 pp 181-273. MUSSO, M.F., & HARTER, M.R. (1975) Visually evoked potentials and selective masking with patterned flashes of different spatial frequencies. Vision Res. 15: 231-238. NAATANEN, R. (1969) Anticipation of relevant stimuli and evoked potentials: A comment on Donchin's and Cohen's "averaged evoked potentials and intramodality selective attention". Perceptual & Motor Skills 28: 639-646. PICTON, T.W., & HILLYARD, S.A. (1974) Human auditory evoked potentials. II. Effects of attention. Electroenceph. clin. Neurophysiol. 36: 191-200 PICTON, T.W., HILLYARD, S.A. GALAMBOS, R. (1973) Cortical evoked responses to omitted stimuli, in Major Problems of Brain Electrophysiology. Ed. M. N. Livanov, USSR Academy of Sciences. PORJESZ, B. & BEGLEITER, H. (1975) The effects of stimulus expectancy on evoked brain potentials. Psychophysiology 12: 152-157. PURVES, S.J., & Lav, M.D. (1975) Visual evoked response to patterned stimulation. Can. J. Neurol. Science 2: 341. REGAN, D. (1975) Recent advances in electrical recording from the human brain. Nature 253: 401-407. REGAN, D. (1974) Fourier analysed EPs in studies of multiple sclerosis, brain lesions and ocular refraction. Paper presented at the International Symposium on Cerebral Evoked Potentials in Man. Brussels, Belgium. REGAN, D. (1972) Evoked Potentials in Psychology, Sensory Physiology and Clinical Medicine. Chapman and Hall (London). RIETVELDn W.J., TORDOIR, W.E.M., & DUYFF, J.W. (1965) Contribution of fovea and parafovea to the visual evoked response. Acta Physiol. Pharmacol. Neerl. 13: 330-339. RIFTVELD, W.J., TORDOIR, W.E.M., HAGENOUW, J.R.B., LUBBERS, J.A. & SPOOR, T.A (1967) Visual evoked responses to blank and to checkerboard patterned flashes. Acta Physiol. Pharmacol. Neerl. 14: 259-285. RITTER, W., SIMSON, R. & VAUGHAN, H.G. (1974) Topographic analysis of task-related cerebral potentials. Paper presented at the International Symposium on Cerebral Evoked Potentials in Man. Brussels, Belgium. RITTER, W. & VAUGHAN, H.G. (1969) Average evoked responses in vigilance and discrimination: A reassessment. Science 164: 326-328. ROHRBAUGH, J., DONCHIN, E. & ERIKSEN, C. (1974) Decision making and the P300 component of the cortical evoked response. Perception & Psychophysics 15: 368-374. RUCHKIN, D.S. & JOHN, E.R. (1966) Evoked potential correlates of generalization. Science 153: 209-211. RUCHKIN, D.S. & SUTTON, S. (1973) Visual evoked and emitted potentials and stimulus significance. Bull. Psychon. Soc. 2 (3) 144-146. RUCHKIN, D.S. (1974) Latency characteristics and t r i a l by t r i a l variation of emitted potentials. Paper presented at the International Symposium on Cerebral Evoked Potentials in Man. Brussels, Belgium. SHIPLEY, T., JONES, R.W. & FRY, A. (1965) Evoked visual potentials and human color vision. Science 150: 162-164. SPEHLMANN, RAINER (1965) The averaged e l e c t r i c a l responses to diffuse and to patterned ligh t i n the human. Electroenceph. c l i n . Neurophysiol. 19: 560-569. SQUIRES, K., HILLYARD, S., & LINDSAY, P. (1973) Vertex potentials evoked during auditory signal detection: Relation to decision c r i t e r i a . Perception & Psychophysics 14: 265-272. SQUIRES, N.K., SQUIRES, K.C. & HILLYARD, S.A. (1975a) Two varieties of long latency positive waves evoked by unpredictable auditory stimuli in man. Electroenceph. c l i n . Neurophysiol. 38: 387-402. SQUIRES, K.C., SQUIRES, N.K. & HILLYARD, S.A. (1975b) Decision-related c o r t i c a l potentials during an auditory signal detection task with cued observation intervals. J . of Exp. Psych.: Human perception and performance 1: 268-279. SUTTON, S., BRAREN, M., ZUBIN, J., & JOHN, E.R. (1965) Evoked-potential correlates of stimulus uncertainty. Science 150: 1187-1188. SUTTON, S., RUCHKIN, D.S. & TUETING, P. (1974) "I d i s t i n c t l y heard the clock not strike": the problem of the physiological response to the non-occurrence of stimuli. Paper presented at the International Symposium on Cerebral Evoked Potentials i n Man. Brussels, Belgium. SUTTON, S., TUETING, P., ZUBIN, J., & JOHN, E.R. (1967) Information delivery and the sensory evoked potential. Science 155: 1436-1439. TEPAS, D.I., GUTERAS, V.L., KLINGMAN, R.L. (1974) V a r i a b i l i t y of the human averaged evoked response to visual stimulation. A Warning. Electroenceph. c l i n . Neurophysiol. 36: 533-537. TUETING, P., SUTTON, S., & ZUBIN, J. (1971) Quantitative evoked potential correlates of the probability of events. Psychophysiology 7: 385-394. TWEEL, L.H. VAN DER & VERDUYNLUNEL, H.F.E. (1965) Human visual responses to sinusoidally modulated l i g h t . Electroenceph. c l i n . Neurophysiol. 18: 587-598. VAUGHAN, H.G. Jr. (1966) The perceptual and physiologic significance of visual evoked responses recorded from the scalp i n man. Vision Res. 6; Suppl 1; 203-223. WALTER, D.O., RHODES, J.M. & ADEY, W.R. (1967) Discriminating among states of consciousness by EEG measurements. A study of four subjects. Electroenceph. c l i n . Neurophysiol. 22: 22-29. 117. WALTER, W.G. (1964) The convergence and interaction of visual, auditory, and t a c t i l e responses i n human non-specific cortex. Ann. N.Y. Acad. Sci . 112: 320-361. WEINBERG, H., WALTER, W.G. & CROW, H.J. (1970) Intracerebral events i n humans related to real and imaginary stimuli. Electroenceph. c l i n . Neurophysiol. 29: 1-9. WHITE, C.T., & EASON, R.G. (1966) Evoked c o r t i c a l potentials i n relation to certain aspects of visual perception. Psycholog. Monogf. 80, No. 24. WOOD, C.C., GOFF, W.R. & DAY, R.S. (1971) Auditory evoked potentials during speech perception. Science 173: 1248-1251. YOSHIDA, S., IWAHARA, S., NAGAMURA, N. (1975) The effect of stimulus orientation on the visual evoked potential in human subjects. Electroenceph. c l i n . Neurophysiol. 39: 53-57. Addendum: BAHN, A. (1972) Basic Medical S t a t i s t i c s . Grune & Stratton New York pp. 39-41. SCHEFFE, H. (1959) Analysis of Variance. John Wiley & Sons. APPENDIX A Computer Hardware and Data Acquisition Program (VER03) - PDP 11/20 with 2OK core and Extended Arithmetic element operating under a modification of DOS-11. - Timing controlled by a 100 Hz programmable clock (KWll-P) - Projector control for presentation of stimuli, monitoring subject's response is done by a DRll-A general purpose interface. - A/D conversation is done by a 32-channel, programmable-gain, bipolar AD01-D. Conversion time is 22 ys/channel. - Operator conversational terminal is a VT05-C. - Intermediate data storage: 256Kbyte RF/RSll disk. 8ys/byte transfer rate. - Long-term data storage: 289Kbyte TU56 Dectape and Digi-data model 1700 9 track magtape. - On-line display of data: VT01 (Tektronix 611) storage display. - Off-line display of data: Matrix 100A printer/plotter (Versatek). Core requirements: Data are stored as 8-bit bytes. One t r i a l generates 896 words. 120 trials generate 1680.0 64 word blocks on RFll disk. Programs to control the VER experiments are written in assembler (MACRO-11) and occupy 3 1/2 K words of core. There is 5K additional allotted for core data buffers, plus 3K for the resident DOS monitor serving i/0 functions. Figure 19 shows a flow chart of the VER03 program for paradigm 1. 119 VER03 ^ Initial dialogue: name, s to be stored as part of header for subsequent data. optionally set limits to control trial rejection; set flag which may discard and proceed or else display trial and invite operator action. select paradigm # to be run. whether or not to clear out previous disk data in the relevant area. display trial on VT01 and enter a conver-sational mode wi th operator. One possibility is that op. decides to delete or revise imi ts. move the collected! data from disk to dectape enter conversational mode. Leave only when COntinue" indicated. . 19 Flow chart indicated the main features of the VEB03 Program 120.. APPENDIX B. Application of Stepwise Discriminant Analysis to evoked Response Studies. Multivariate S t a t i s t i c s Discriminant analysis i s based on multivariate s t a t i s t i c a l theories and the evoked potential must be considered as a multivariate observation for this analytical technique to be applicable. For this conceptualization the voltage values at successive time points at which measurements are made are considered as different variables and each variable as the sum of the noise plus the event-related a c t i v i t y . The average evoked potential can be considered as the estimate of the mean vector of a multivariate normal distribution. Hypotheses about such estimates can be tested using the methods of multivariate s t a t i s t i c a l analysis. The assumptions about the EEG data that this model requires are discussed by Donchin (1966) and Walter, Rhodes & Adey (1967). These include: 1. the c o r t i c a l response i s independent of the "noise" process (ongoing EEG) 2. the data must be obtained by a process or random sampling, i.e. the sample taken during one stimulus replication must be independent of the one taken during the next replication. 3. the process generating the distribution i s stationary over the presentations, i.e. the parameters of the distribution of the noise and of the response processes do not change from one presentation of a given stimulus to the next. Donchin & Herning (1971) assume that the model i s appropriately descriptive 121. of the data and then show that SWDA provides some useful insights into the data provided by a simulation study. Walter et a l . (1967) also mention that although the various normality assumptions are violated to seme extent by EEG data, they s t i l l consider that the method selects parameters that are useful for further testing. DISCRIMIISIANT ANALYSIS Discriminant analysis i s a cl a s s i f i c a t i o n technique that uses data obtained from members of different groups whose group memebership i s known, to derive c r i t e r i a for the cl a s s i f i c a t i o n of an observation whose group membership i s doubtful. The discriminant analysis program used i n this study and a l l others referred to i n this dissertation i s the Stepwise Discriminant Analysis Program (SWDA) available as part of the UCLA BMD program package (Dixon, 1965, Program 07M) . Its use i s particularly advantageous i n evoked response work because i t takes into account the considerable correlation that exists between successive time points i n the averaged evoked response. The program i n i t i a l l y computes an F ratio for a l l of the variables entered and then selects the variable with the highesr ratio, which essentially means the variable with the largest inter-group variance as compared to i t s within-group variance. The program then computes the conditional probabilities for a l l of the variables related to the one i t has just chosen, and reduces the F ratios of those that have a high correlation with i t (essentially this prevents the time points adjacent to the one just chosen, which are presumably part of the same component of the waveform, from being selected). Then i n the second 'step' of the program the,next variable with 122. the highest F ratio i s chosen and the above process of adjusting the remaining F ratios i s repeated. Thus a second point i s selected that i n combination with the f i r s t point produces maximal improvement i n the discrimination. The steps continue u n t i l there are no points whose F ratio i s higher than some preset l i m i t (.2 i n this study), or u n t i l the number of steps designated in the program instruction i s reached (which was 6 or 7 i n this study). At the end of each step the program also computes a discriminant function based on the variables i t has so far selected. This i s a mathematical function that can be evaluated for given values of the variables included i n i t . Evaluation of the function for any set of variables results i n a prediction as to which of the groups this set of variables was most l i k e l y to have come from. After the variables are selected and the discriminant function produced, the BMD:07M program computes some s t a t i s t i c s that indicate the efficiency of the discrimination. One such value i s the U value, which ranges frcm 1.00 to 0.00. The smaller the U value, the larger the difference between the groups. The program also produces a c l a s s i f i c a t i o n matrix that indicates the number of the input observations that are c l a s s i f i e d into each of the original groups when the value of the discriminant function i s computed for each of these observations, and i t t a l l i e s the number of correct classifications and expresses this number as a percentage of the total number of observations. However, i t should be noted that this i s an a p o s t e r i o r i c l a s s i f i c a t i o n . Even i f a high percentage of the original data i s correctly c l a s s i f i e d , i t does not necessarily indicate that the function 123. will perform well (i.e. discriminate well) on new data; i t must be tested. The evaluation of the discriminant function for new observations is performed by another program called UBC:CLASS. Finally i t may be pointed out that the use of this method is not a small undertaking: approximately 1.6 million digital voltage values (for occipital channels only) were transformed to 420 thousand points for input to the BMD:07M and UBC:CLASS programs (67 points for each single trial) that computed and tested the discriminant functions reported in this study. 124. APPENDIX C. Terras from experimental psychology. Psychological variables The term i s synonymous with psychological constructs and generally refers to those elements having to do with the state of the subject (relaxed, attentive, anxious, expectant etc.) and the context within which a stimulus i s delivered. The context i s related to the task assigned to the subject; for example whether or not a stimulus i s relevant depends on what the subject has been told to do. Relevance can also be assigned by associating a stimulus with an unconditioned stimulus, without separately instructing the subject, and frequently both methods are used to ensure relevance as i n paradigm 2 of the experimental work presented i n Chapters II and TV. Stimulus Meaning: Simple sensory stimuli such as cl i c k s , flashes or e l e c t r i c shocks have no inherent meaning. Some complex stimuli do have inherent meaning, in the sense that, given common cultural and developmental backgrounds of the subjects, they w i l l convey the same information or w i l l be symbolic of the same things to a l l subjects (letters of the alphabet, numbers and concrete nouns are ccmmon examples of such stimuli). The geometric shapes used i n this study are complex stimuli. The square and c i r c l e are familiar patterns and can be easily labelled by ccmmon concrete nouns by the subject. The e l and omega shapes are not familiar patterns to the subject and thus did not have inherent meaning and cannot be as easily labelled by subjects in our culture. 125. In p h y s i o l o g i c a l and experimental psychology, meaning can be a s s i g n e d t o any s t i m u l u s a c c o r d i n g e i t h e r t o the e x p e r i m e n t a l d e s i g n ( e . g . c o n d i t i o n e d s t i m u l u s - u n c o n d i t i o n e d s t i m u l u s p a i r i n g s ) o r t o the i n s t r u c t i o n s g i v e n by the experimenter ( e . g . s t i m u l u s appearance s i g n i f i e s the s u b j e c t must p r e s s a button) o r t o a combination o f both d e s i g n and i n s t r u c t i o n s . 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0093878/manifest

Comment

Related Items