UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Closed-loop control versus preprogrammed control in a self-paced and ballistic response Roy, Eric Alexander 1973

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

Item Metadata

Download

Media
831-UBC_1973_A7_5 R69.pdf [ 5.47MB ]
Metadata
JSON: 831-1.0077223.json
JSON-LD: 831-1.0077223-ld.json
RDF/XML (Pretty): 831-1.0077223-rdf.xml
RDF/JSON: 831-1.0077223-rdf.json
Turtle: 831-1.0077223-turtle.txt
N-Triples: 831-1.0077223-rdf-ntriples.txt
Original Record: 831-1.0077223-source.json
Full Text
831-1.0077223-fulltext.txt
Citation
831-1.0077223.ris

Full Text

CLOSED-LOOP CONTROL VERSUS PREPROGRAMMED CONTROL IN A SELF-PACED AND BALLISTIC RESPONSE by ERIC ALEXANDER ROY B.Sc, University of Waterloo, 1971 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENT FOR THE DEGREE OF MASTER OF PHYSICAL EDUCATION in the School of Physical Education and Recreation We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA April, 1973 In present ing th is thes is in p a r t i a l f u l f i l m e n t o f the requirements fo 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 sha l l make i t f r e e l y a v a i l a b l e for reference and study. I fu r ther agree that permission for extensive copying o f th is 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 is representa t ives . It is understood that copying or p u b l i c a t i o n o f th is thes is f o r f i n a n c i a l gain sha l l not be allowed without my wr i t ten permiss ion . Department of Physical Education The U n i v e r s i t y o f B r i t i s h Columbia Vancouver 8, Canada Date A P r 1 1 18> 1 9 7 3 ABSTRACT The main purpose of this study was to determine the generality of the closed-loop theory and the preprogramming theory as an explanation for the learning and maintenance of performance in a highly practiced self-paced and ballistic response. The methodology used to investigate this problem involved comparing performance, following the withdrawal of knowledge of results, under changed or interrupted feedback conditions to a control condition in which feedback was the same as that in acquisition. Subsidiary problems involved 1) examining the effects of changing or interrupting feedback during KR withdrawal following low practice in acquisition and, 2) examining the differential effects of low and high practice in acquisition on performance in each response type during KR withdrawal under each of the three feedback conditions. The experimental task involved learning to move a cursor on a track from one end of the track to the other in 1.0 seconds. Two types of responses were used: 1) self-paced, in which the subject was permitted to hold on to the cursor for the entire length of the track and, 2) ballistic, in which the subject had to release the cursor after he moved i t only about one sixth of the track distance. Sixty students of the University of British Columbia served as subjects. The results indicated that the preprogramming theory explained the learning and maintenance of performance in a highly practiced ballistic response, while the closed-loop theory was most applicable to the highly practiced self-paced response. Secondly, after 15 trials of practice in acquisition both response types were dependent on feedback, but the amount of feedback necessary was much less in the ballistic response than in the self-paced response. Thirdly, in the ballistic response, a comparison of performance in KR withdrawal following a small amount of practice in acquisition with that following a large amount of practice indicated that there was a transition from a primitive preprogramming mechanism which was somewhat dependent on feedback to a well developed preprogrammed mechanism which was not dependent on feedback. Finally, a closed-loop mechanism was suggested for the self-paced response following both small and large amounts of practice in acquisition. ACKNOWLEDGEMENTS I wish to express my appreciation to the members of my committee, Dr. R. G. Marteniuk, Dr. R. W. Schutz, Dr. W. G. Davenport and Dr. D. Foth, for their encouragement and comments throughout the preparation of this thesis. To my chairman, Dr. Marteniuk, I wish to express a special appreciation for his guidance throughout my academic career. Finally, to my wife, who typedthis thesis, I wish to express my deepest gratitude for her patience and understanding. TABLE OF CONTENTS Page LIST OF TABLES iv LIST OF FIGURES v CHAPTER I STATEMENT OF THE PROBLEM 1 Introduction 1 Statement of the Problem 6 Subproblems 6 Hypotheses 6 Definition of Terms 9 Delimitations 10 Assumptions and Limitations 10 Significance of the Study 11 II A CRITICAL REVIEW OF THE LITERATURE 13 Introduction 13 Feedback in Performance 14 Conditioning Theories 14 Teleological Theories 22 Preprogramming in Motor Performance 34 Lashley - On Central Control of Performance 34 Reduction of Feedback 35 Efference in Perception 39 Other Evidence for Preprogramming 41 Control Mechanisms in Preprogramming 43 Discussion 45 III METHODS AND PROCEDURES 49 Subjects 49 Apparatus 49 Experimental Design 53 Procedures 54 Experimental Conditions 56 Analysis of the Data 60 IV RESULTS AND DISCUSSION 62 Results 62 Analysis of Constant Error and Variable Error 62 Analysis of Intra-Track Time 75 Performance on Secondary Task 78 Summary of Main Findings 79 Discussion 80 i i i V SUMMARY AND CONCLUSIONS 93 Summary 93 Conclusions 95 BIBLIOGRAPHY 97 APPENDICES A. Apparatus 104 B. Statistical Analysis 107 C. Tables of Means 115 IV LIST OF TABLES Table Page I Homogeneity and Normality of Error Variance in Data for CE and VE Measures 63 II The Main Effects in Constant Error 67 III The Main Effects in Variable Error 71 IV The Main Effects in the Analysis of Intra-Track Times 75 V The Main Effects in the Analysis of Variability (SD) of Intra-Track Times 78 VI Per Cent Correct Recall on the Secondary Task 78 LIST OF FIGURES Figure Page 1 Feedback and Learning Effects on the SP and B Responses in CE 68 2 The Blocks Effect at Each Level of Learning 70' 3 Feedback and Learning Effects on the SP and B Responses in VE 73 4. The Effect of Type of Feedback on the Intra-Track Times in an SP Response 77 CHAPTER I STATEMENT OF THE PROBLEM Introduction The availability of information about performance is essential for the learning of a motor response (Bilodeau, 1969; Adams, 1971). This information is provided extrinsically, intrinsically or centrally (Keele, 1968; Greenwald, 1970). Extrinsic information is provided by knowledge of results (KR). Intrinsic information is response-produced and can be either interoceptive (eg. proprioception) or exteroceptive (eg. vision, audition). Central information involves monitoring the efferent outflow from an ordered response (Taub, Bacon and Berman, 1965; Laszlo and Manning, 1970). The use of extrinsic and intrinsic information in motor performance has been incorporated into models of the human operator as a component in a control system (Craik, 1947, 1948), into models describing man as an engineering system (Chase, 1965a, 1965b; Welford, 1968; Craik, 1970), into closed-loop models (Miller, Gallanter and Pribram, 1960; Adams, 1971) and into conditioning models (Greenwald, 1970). Central information about motor performance has been described in efferent readiness theory (Festinger and Canon, 1965; Festinger, Ono, Clarke and Bamber, 1967), reafference theory (Held, 1961), and motor programming theory (Keele, 1968; Laszlo and Manning, 1970). Two models which are of interest in this study are closed-loop learning theory (Adams, 1971) and preprogramming theory (Keele, 1968; 2 Laszlo and Manning, 1970). Both theories provide an explanation for learning and for the maintenance of learned performance following the withdrawal of KR. Closed-Loop Theory. In the closed-loop learning theory, during acquisition, KR is used to develop a memory trace while KR and peripheral feedback develop a perceptual trace. The perceptual trace acts as a reference mechanism (standard) and is the subject's representation of the correct response. As learning progresses the perceptual trace becomes progressively more precise. Indeed, Adams, Goetz and Marshall (1972) demonstrated that the greater the amount of practice in acquisition the more well defined was the perceptual trace. Following the withdrawal of KR, the perceptual trace becomes the performance control mechanism in that peripheral feedback is compared to i t in order to detect error: the subject continues to respond until there is a match between the perceptual trace and the feedback at which time he recognizes the error to be minimal. Further, Adams, Goetz and Marshall (1972) indicated that the perceptual trace developed during acquisition was specific to the type of feedback used during that period. Consequently, i f the characteristics of the feedback were altered following the withdrawal of KR from those which prevailed during acquisition, performance deteriorated. The memory trace is used as a mechanism for selecting and initiating a movement. Adams (1971) considers the memory trace to be a type of motor program. However, unlike the motor program in preprogramming theory, its importance in performance control is hypothesized to be minimal. 3 Preprogramming Theory. In the preprogramming theory KR is used, during acquisition, to develop a motor program while a standard is developed by KR in conjunction with peripheral feedback. Two hypotheses concerning the maintenance of learned performance following acquisition are held. The firs t states that performance is primarily under the control of the motor program (Keele, 1968). The other hypothesis states that a central loop established during acquisition between the motor program and the standard is used to indicate to the subject the accuracy of his response (Laszlo and Manning, 1970). The subject compares the efferent command elicited by the motor program to the standard for error detection. It should be noted that these hypotheses possess the common tenet that feedback is not required for performance control following the withdrawal of KR. The similarities between the closed-loop and preprogramming theories are (1) they both require the presence of KR and intrinsic feedback during acquisition; (2) during acquisition* a type of standard is hypothesized to develop in both; (3) some type of control loop is involved in order to detect error following KR withdrawal. The major difference between the theories which is of interest in this study, appears to be in regard to the location of the control loop during KR withdrawal. In motor programming theory the loop is regarded as being central involving a comparison of efferent information from the motor program with the standard (Laszlo and Manning, 1970). In closed-loop theory the loop is peripheral involving intrinsic feedback and the perceptual trace (Adams, 1971). This difference further suggests that there is a differential requirement for feedback. In closed-loop theory intrinsic feedback is essential to performance control during this period while intrinsic feedback is not necessary in preprogramming theory. 4 Since the major difference between the two theories appears to be in regard to the necessity of intrinsic feedback for control during KR withdrawal, i t seems that the theory used to expalin performance control may be dependent on the availability of feedback during the execution of the response. If a response is such that feedback cannot be used for guidance during execution, there may be tendency to develop preprogrammed control during acquisition. Alternately, i f feedback is available for guidance, a closed-loop mechanism based on feedback may develop. Some recent evidence by Schmidt and Russell (1972) bears directly on this problem. These authors were investigating the relationship between the movement time (MT) of a response and the degree of preprogramming developed. They found that, regardless of movement velocity, the shorter the MT the greater the degree of preprogramming as indicated by the index of preprogramming (Schmidt, 1972). Responses which involved an MT of .150 seconds or less were completely preprogrammed while responses with an MT greater than .750 seconds involved much less preprogramming and more atten-tion to feedback. From this evidence i t follows that in a response which is fast (MT<.150 seconds), thus preventing the use of feedback for control, a preprogramming will result while a response which is slower (>1 second), may involve a closed-loop mechanism since feedback is available to guide per-formance. Two major classifications of responses which seem to exhibit this difference in the availability of feedback for control are self-paced and ballistic responses. A self-paced response is one which is primarily guided by intrinsic feedback and characterized by a relatively long MT. A ballistic response is one in which feedback cannot be used to guide the response and is characterized by a relatively short MT due to the speed with which i t is executed (Adams, 1971). Thus, i t is suggested that a 5 self-paced response would be more likely to develop closed-loop control during acquisition since the longer MT allows for the use of feedback during execution. On the other hand, a ballistic response may show a greater tendency to develop preprogrammed control during acquisition due to the shorter MT. In either case the KR withdrawal phase in the performance of the response is the period in which i t is possible to delineate which mechanism is controlling performance. This delineation results from a consideration of the differential role of feedback in each theory. Since feedback is proposed to play l i t t l e role in a mechanism involving programmed control, manipulation of feedback by either preventing its use or changing its characteristics from those which prevailed during acquisition should have l i t t l e effect. On the other hand, manipulation of feedback by either changing its characteristics during KR withdrawal from those present during acquisition (Adams, Goetz and Marshall, 1972) or by reducing feedback (Chase, Sutton and Rapin, 1961), have been shown to have a detrimental effect on the performance of a response under closed-loop control. It is the purpose of this study, then, to attempt to determine what control mechanism is involved in the two types of responses. The experi-mental technique to be used to determine which of these theories best explains learning and maintenance of performance in each response involves comparing performance, following the withdrawal of KR, under changed or interrupted feedback conditions with a control condition in which feedback is the same as that used in acquisition. Small differences between the changed or interrupted feedback conditions and the control condition would suggest programmed control while large differences would suggest closed-loop control. 6 Statement of the Problem The purpose of this study is to investigate the preprogramming theory and the closed-loop theory of learning as an explanation for the learning and maintenance of performance in highly practiced ballistic and self-paced responses. Subproblems The subproblems are: 1. To determine the effects of changed feedback from that which was present in learning or of interrupting feedback on the performance of a ballistic and self-paced response under low practice during K R withdrawal. 2. To determine for each response type the effects of low and high practice in acquisition on the performance following the withdrawal of K R when feedback is the same as that in acquisition, when feedback is different from that used in acquisition, and when feedback is interrupted. Hypotheses The hypotheses are: 1. The interruption of peripheral feedback or changing the characteristics of feedback from what was used in learning results in a change in constant error (CE) and variable error (VE) relative to the 7 unchanged feedback cond i t ion in a h igh ly p r a c t i c e d s e l f - p a c e d response on KR withdrawal t r i a l s . No such change occurs in a h igh ly p r a c t i c e d b a l l i s t i c response. In the s e l f - p a c e d task the perceptual t r a c e , which exerts performance cont ro l f o l l o w i n g a c q u i s i t i o n , requi res per iphera l feedback. F u r t h e r , t h i s feedback must be the same as that used during a c q u i s i t i o n . In the b a l l i s t i c response, however, s i n c e i t i s hypothesized that a preprogramming of the response o c c u r s , cent ra l cont ro l o f the response without re ference to feedback i s used. Thus a l t e r i n g or prevent ing the use of per iphera l f e e d -back should have no e f f e c t . 2. With low p r a c t i c e in a c q u i s i t i o n , changing feedback or i n t e r r u p t i n g feedback dur ing KR withdrawal r e s u l t s i n no change in CE and VE r e l a t i v e to the unchanged cond i t ion i n both a s e l f - p a c e d and b a l l i s t i c response as a l l feedback cond i t ions r e s u l t in equa l ly poor performance. A great deal o f p r a c t i c e in a c q u i s i t i o n should enable the develop-ment o f the cont ro l mechanism which i s subsequently used to cont ro l performance fo l low ing the withdrawal of KR. S ince i t had been hypothesized that a c l o s e d - l o o p mechanism dependent on feedback develops in the s e l f -paced response whi le a preprogramming mechanism not dependent on feedback was hypothesized to develop in a h igh ly p r a c t i c e d b a l l i s t i c response , then , as pred ic ted in hypothesis one, there should be a large e f f e c t of manipu-l a t i n g feedback in KR withdrawal f o r a s e l f - p a c e d response but not f o r a b a l l i s t i c response. When only a small amount of p r a c t i c e is g i v e n , however, the cont ro l mechanism in e i t h e r response would not have enough time to develop. T h e r e f o r e , the mechanisms in both response types should s t i l l be 8 dependent on KR as they have not developed to a sufficient degree to operate independently of KR. Consequently, withdrawing KR should result in an approximately equal performance decrement in all feedback conditions under both response types. 3. For the self-paced response high practice in acquisition results in significantly lower VE and CE than low practice in KR withdrawal only when feedback is the same as that used in learning. For the self-paced response when feedback is manipulated during KR withdrawal by either changing its characteristics or preventing its use the high practice and low practice conditions should approach each other in terms of CE and VE because, in hypotheses one and two, a decrement in both learning conditions was predicted when feedback is manipulated. When feedback during KR withdrawal is the same as that in acquisition, however, the high practice condition should have significantly lower CE and VE than the low practice condition because a decrement in low practice only (hypo-thesis two) was predicted in this feedback condition. 4. For the ballistic response high practice in acquisition results in significantly lower VE and CE than low practice in KR withdrawal regard-less of changes or interruption of feedback. For the ballistic response, since a decrement in performance was predicted under all feedback conditions in low practice (hypothesis two) while no change in performance was predicted in high practice (hypothesis one), then the high practice condition should have consistently lower CE and VE than the low practice condition. 9 D e f i n i t i o n of Terms Perceptual T r a c e . The perceptual t race is a memory s ta te which represents the re ference mechanism or standard to which the sub ject compares per iphera l feedback in order to detect e r r o r on t r i a l s in which external KR i s not present . This t race develops during a c q u i s i t i o n of a motor response and is formed by the a s s o c i a t i o n of KR and per iphera l response-produced feedback (Adams, 1971). Memory T r a c e . The memory t race i s a memory s t a t e which represents the c o r r e c t response and i s the mechanism f o r s e l e c t i n g and i n i t i a t i n g a movement response. Th is t race develops in a c q u i s i t i o n as a funct ion of p r a c t i s i n g the c o r r e c t response (Adams, 1971). Motor Program or Preprogrammed Mechanism. A motor program i s a preprogrammed sequence of motor responses that can be executed without re ference to per iphera l feedback (Kee le , 1968). I n t r i n s i c Feedback. I n t r i n s i c feedback i s a source of informat ion which the sub jec t acquires through the a f f e r e n t sensory channels as a r e s u l t of h is own movement (Greenwald, 1970). I n t r i n s i c feedback in t h i s study i s aud i tory and k i n e s t h e t i c . Audi tory feedback comes from the sound of the response whi le k i n e s t h e t i c feedback comes from the f e e l of the response. E f f e r e n c e . E f ference re fe rs to the motor outf low or e f fe ren t command e l i c i t e d by the CNS. Monitor ing e f fe ren t outf low r e f e r s to the process by which the subject attends to or monitors e f fe rence (Fes t inger and Canon, 1965; F e s t i n g e r , Ono, Clarke and Bamber, 1967). 10 Knowledge of Results (KR) or E x t r i n s i c Feedback. Knowledge of r e s u l t s i s an external source of informat ion which the sub jec t uses to learn a motor response. This informat ion can be provided in any sense. Quant i ta t i ve KR involves informing the subject of both the magnitude and d i r e c t i o n of his e r r o r (B i lodeau , 1969). S e l f - p a c e d Response. A s e l f - p a c e d response i s one in which the subject responds at h is own pace (Adams, 1971). B a l l i s t i c Response. A b a l l i s t i c response i s one in which the subject must move very q u i c k l y (movement time - .150 s e c . ) Thus , the accuracy of the response i s determined by the i n i t i a l impulse (G ibbs , 1954; Schmidt and R u s s e l l , 1972). Del imi ta t ions 1. The study i s de l im i ted to a t iming response by the method of p roduc t ion . 2. The study i s de l im i ted to one t iming response of one second. Assumptions and L imi ta t ions The fo l low ing assumptions are made: 1. In a s e l f - p a c e d response i n t r i n s i c feedback i s used during the execut ion o f the task f o r cont ro l of performance (Posner and Kee le , 1969). n 2. In a ballistic response intrinsic feedback is not used for performance control. 3. The diversion of attention from feedback in a self-paced response will prevent its use in guiding a response without affecting the efferent commands for movement (Posner, 1969). 4. A decrement in performance on non KR trials due to the withdrawal or changing of intrinsic feedback indicates that intrinsic feedback is required for accurate task performance (Adams, Goetz and Marshall, 1972). The investigation is limited by: 1. The accuracy of the timing equipment. 2. The sample size of sixty subjects. Significance of the Study At present there appears to be some controversy as to whether the motor program theory or the closed-loop theory provide the most appropriate explanation for the maintenance of performance following the withdrawal of KR. This controversy may be resolved when considering that the theory used may depend on the type, of response employed. The motor program theory may provide greatest explanatory power in a response which cannot be guided by peripheral feedback, ie., a ballistic 12 response. A preprogramming of the b a l l i s t i c response may be r e q u i r e d . C losed- loop theory may be used to exp la in the maintenance of performance fo l low ing a c q u i s i t i o n in a response which i s p r i m a r i l y guided by per iphera l feedback, i e . , s e l f - p a c e d response. Thus, the j u s t i f i c a t i o n f o r th is study i s that i t i s an attempt to a s c e r t a i n the g e n e r a l i t y of these two theor ies in exp la in ing the maintenance o f learned performance. CHAPTER II CRITICAL REVIEW OF THE LITERATURE Introduction The control of a motor response following the withdrawal of external KR can involve the use of intrinsic feedback, a peripheral source of infor-mation, or i t can involve a more central source of information based on monitoring the motor command. Mechanisms based on a peripheral source of information are characterized by the necessity for feedback (Adams, 1968; Keele, 1968; Greenwald, 1970). Many theories have been proposed for how this intrinsic feedback is used in learning and subsequent control of motor performance. These theories will be reviewed in some detail, with special emphasis placed on a description of the closed-loop theory of motor learning (Adams, 1971). The other type of mechanism alluded to above is based on a more central source of information. These central mechanisms may involve either a preprogramming of the response based on efferent outflow (Festinger, Ono, Clarke and Bamber, 1967; Keele, 1968) or some type of central control loop (Taub, Bacon and Berman, 1965; Laszlo and Manning, 1970). These central mechanisms are characterized by the fact the intrinsic feedback from the periphery is not necessary for control. 13 Feedback in Performance The concept of feedback in performance stems from two primary sources. The fir s t source views feedback in the context of stimulus-response (S-R) learning. In this review this theoretical approach will be referred to as the conditioning theory of feedback. The other perspective on feed-back views man as "an information processor who compares incoming sensory feedback from responses with a stored representation of what feedback from correct performance should be". (Greenwald, 1970: 79) This approach will be referred to as the teleological theory of feedback since i t represents man's capability for purposeful action in terms of error detection and correction. Conditioning Theories Response Chaining. One of the earliest theoretical accounts of the use of response-produced feedback was serial response chaining (James, 1890a; Washburn, 1916; Hull, 1930; Watson, 1930). Essentially the theory involved the substitution of response-produced stimuli for situational stimuli in the control of movement. When initially performing a task performance was seen as being primarily guided by situational stimuli which were extrinsic to performance, such as KR. As performance became more well learned each response was consistently preceded, not only by its particular situational stimuli, but also by intrinsic stimuli from preceding movements. These intrinsic stimuli, whether they be exteroceptive (eg. vision) or intero-ceptive (proprioception), share the attribute of being contingent on the performer's own behaviour. Through S-R contiguity i t was thought that control of performance was transferred from situational stimuli to intrinsic stimuli. That is, while in original learning and performance each response 14 15 was conditioned to particular situational stimuli, in later performance the response became conditioned to particular intrinsic stimuli. Indeed, Washburn defined a movement system as "a combination of movements so linked together that the stimulus furnished by the actual performance of certain movements is required to bring about other movements". (Washburn, 1916: 11) Greenwald (1970) in discussing this type of motor control described what kind of experiment would be necessary to demonstrate the existence of serial chaining. The experiment would involve three steps: (a) the subject would have to learn a skill involving a series of responses originally under the control of a series of situational stimuli; (b) the skill would be tested following interruption of the sensory pathways for receiving the situational stimuli while leaving intact the pathways for intrinsic feedback; (c) the skill would again be tested with sensory pathways for both situational and intrinsic feedback blocked. The purpose of this particular experimental paradigm would be to demonstrate that the intrinsic feedback stimuli acquired the ability to guide performance. Evidence to this effect would be found i f performance in step b was superior to that in step c. Research reported by Adams (1968) on animal studies tends to support the response chaining hypothesis. However, Greenwald (1970) criticized much of this work since the researchers did not attempt to utilize the experi-mental paradigm suggested above. The situational and intrinsic stimuli are confounded unless careful consideration is given to exactly what senses are being used to receive situational stimuli. Indeed, Honzik (1936) found 16 that, in learning a maze, rats could use, interchangeably, the senses of vision, audition and olfaction. Consequently, attempting to distinguish which of these stimuli were situational and which were intrinsic proved to be an almost impossible task. One means of sorting out the confusion would be to regard proprioception as the intrinsic feedback. Much research using this tactic, however, proved only to confuse the issue more (see Adams, 1968 pp 487-489) since the workers did not cut out all feedback except intrinsic (proprioceptive) feedback as was suggested in step b of the Greenwald paradigm. In a reaction to this work reviewed by Adams (1968), Honzik (1936) systematically eliminated various senses, singly and in groups, as well as eliminating, in one group of rats, all senses save proprioception. In the latter group he found that proprioception alone was insufficient for original learning of the maze habit. He never did test the rats using the paradigm suggested by Greenwald; however, he did intimate that proprio-ception could play a role in response chaining (Honzik, 1936: 86). Other evidence for the role of response chaining comes from findings related to the description of overlearned performances, such as speech, by eliminating or altering feedback (Yates, 1963; Smith, 1966). Generally, i t was found that the ability to speak was virtually eliminated i f auditory feedback from the speaker's voice was delayed. In view of the results of this work on the control of routinized, sequential performances, such as speech, strongly implicated the role of response chaining (Greenwald, 1970). Although some evidence supported the idea of response chaining in performance, this process was not necessarily always involved following the 17 elimination of situational stimuli (eg. KR). Studies on the deafferentation of monkeys (Taub and Berman, 1968) demonstrate that monkeys can retain or relearn skills in the absence of any response-produced stimulus information. Further, Lashley (1951) noted that highly skilled performance occurred too rapidly to suggest that stimuli from one response could affect the control of the next response in the sequence. This evidence suggested that the movement patterns have been programmed. Motor programming will be discussed in more detail in the next section. The evidence against response chaining by Lashley (1951) presented above is important when considering tasks which are very fast. This evidence, however, does not negate the existence of response chaining. Greenwald (1970: 77) has suggested that response chaining may be most important in "moderately learned skills requiring repetition of a standard series of movements" or in "a state of transition to centrally organized motor programs". Fractional Anticipatory Goal Theory. Another approach to feedback in conditioning theory was proposed by Hull (1931). He considered feedback as a mediator between each response and the final or goal response. Hull (1930) indicated that any response which was composed of a sequence of component responses could give the subject "foresight". That i s , any part of the total response sequence could be used to anticipate the consequences of the final response or the goal response. Further, any performance which necessitated a sequence of responses could be conditioned to stimuli (eg. drive stimuli) that were present throughout the sequence. These formulations represent a step beyond response chaining because, now, each 18 response is not dependent on the one immediately preceding i t . Thus, Hull used these ideas to explain how unnecessary responses could be dropped out. This process was referred to as short-circuiting by Hull (1930). He applied this idea in formulating the mediating role of sensory feedback from fractional anticipatory goal responses. Hull (1931) stated the fractional portions of the goal response could "short circuit" to the beginning of the response sequence without disturbing the sequence. Thus, the response series could become conditioned to the sensory feedback from the fractional anticipatory goal response. Greenwald (1970) indicated that, with this interpretation of the mediating role of sensory feedback, Hull was able to provide an explanation of purpose in terms of the habit concept. Purpose, in Hullian terms, involved the ability to "respond to a future goal event that is available as a stimulus in the form of sensory feedback from the [anticipated goal response]". (Greenwald, 1970: 77) Hull's formulations require that the subject must be able to anticipate the final or goal response and use sensory feedback from this response to mediate in the performance of responses within the sequence. It is possible, however, that any particular response can be conditioned to its own anticipated feedback. This principle, termed the ideomotor theory, was originally proposed by James (1890b) and has been recently revived by Greenwald (1970). Ideomotor Theory. Essentially, the theory states that a central image composed of various forms of sensory feedback is formed through practice. Through contiguity each response becomes conditioned to its own 19 anticipated feedback represented centrally as the sensory image. When activated this central image or idea results in the corresponding response being produced. In essence, then "an anticipatory image of feedback from an action participates in the selection and initiation of the action". (Greenwald, 1970: 91) Greenwald has discussed several lines of evidence which tend to support the ideomotor principle (see Greenwald, 1970: 89-92). Of primary interest in this discussion, however, is that evidence related to response images. Observational learning studies (eg. Greenwald and Albert, 1968) have indicated that various types of skilled performance are facilitated following visual observation of another's performance. In a visual discrimination avoidance task Greenwald and Albert (1968) found that, just by watching naive subjects learn the task, the observers were able to perform accurately. It appeared that the visual images or representations of visual feedback from performance, developed during observation, subsequently facilitated accurate performance. Although this line of evidence suggested that the image was important in performance, i t did not represent evidence solely in favour of the ideomotor theory since response images are also important in a closed-loop mechanism. The closed-loop mechanism requires that the image be present following the response in order to act as a standard for compari-son with the feedback from the response. The ideomotor mechanism requires that the image be present prior to the response so as to act in response selection. Although these two mechanisms seem somewhat incompatible, i t may be that the image in the ideomotor mechanism is similar to the memory 20 trace in closed-loop theory (Adams, 1971) since the memory trace is purported to be involved in response selection. In summary, the three major conditioning theories are serial response chaining, fractional anticipatory goal responses and ideomotor theory. Response chaining suggests that the integrity of the overall performance is dependent upon the close relationship between each response, and that feedback from preceding responses is used to aid in the performance of the following response. The fractional anticipatory approach, however, considers feedback from the final or goal response as of primary importance while the ideomotor theory regards anticipated feedback of each response as most important. These conditioning theories appear to be deficient in two ways: they tend to be restrictive in their explanatory power and they tend to disregard the purposefulness in human behaviour. The f i r s t criticism especially relates to the fractional anticipatory theory (Hull, 1931). This theory seems to be restricted to instrumental conditioning involving, primarily, rat learning of the maze habit, since i t assumes that the final response is the correct response. In maze learning the final or goal response is the correct response. It is not necessarily true, however, that in human learning tasks the final response is the correct one. The correct response may occur, purely by chance, anywhere in the sequence of responses although the probability of the correct response occurring does increase with practice (Jones, 1962). Nevertheless, i t may be erroneous to state that the final respone in any sequence of responses is the correct one since i t has been shown that learning, in some tasks, can continue 21 even after thousands of trials (Fitts and Posner, 1969). Consequently, in view of this criticism i t seems that this theory may not be very applicable to most human motor tasks. This criticism of restrictiveness can also be levelled at the ideomotor theory since Greenwald seems to have restricted the type of feedback used in conditioning to exteroceptive stimuli such as vision (see Greenwald, 1970: 85). Surely, the performer must be able to go beyond visual control. Indeed, much research shows that visual feedback is only useable when movement is relatively slow (Keele and Posner, 1968). Also, some research shows that control of performance shifts from vision to proprioception as practice continues (Fleishman and Hempel, 1954). There-fore, i t seems that any theory which proposes the use of feedback must include interoceptive stimuli (eg. proprioception) as feedback. The second criticism of conditioning theories involves the observation that these theories tend to view man as a passive link between the stimulus and response not capable of purposeful action. Undoubtedly, this view stems from the foundations of psychology based on the simple reflex arc. Indeed, Adams (1971) writes that despite the emphasis these conditioning theories have on feedback, they are not truly teleological in nature since they are not error centered where feedback is compared against a reference mechanism as a basis for error detection and correction. As will be discussed in the following section, man is not passive; rather he is capable of deciding whether a response is correct based upon the feedback, and, subsequently, capable of correcting this error (Miller, Galanter and Pribram, 1960; Adams, 1971). 22 Teleological Theories Interest in the teleological concept of feedback in performance was revived by Ashby and Rosenbluth (in Miller, Galanter and Pribram, 1960) who observed that machines, involving negative feedback mechanisms, were teleo-logical mechanisms. That i s , they could work to reduce their own error. For many years strict S-R theorists had scoffed at cognitive theorists who postulated mechanisms involving the processing of peripheral feedback (Miller, Galanter and Pribram, 1960). Not until man, himself, was able to build a machine that would do all the things the cognitive theorists had hypothesized were the sceptics willing to entertain thoughts of feedback mechanisms in performance. One of the earliest accounts of this use of feedback was described by John Dewey (1896) as he discussed the servomechanism involved in reaching toward a candle and then jerking back. He recognized that the distinction between the stimulus (feedback from candle heat) and the response (jerking hand back) was a teleological one, ie., one of purpose. The function of the feedback stimulus was to inform of some peripheral occurrence, while the function of the response was to act on this information. Other early accounts of feedback in performance were provided by Troland (1928) who coined the term "retroflex" for sensory feedback. Tolman (1939) related one of the earliest accounts of feedback mechanisms in his "schematic sowbug". Weiner (1948) and Wisdom (1951) developed the "cybernetic hypothesis" stating that the fundamental building block of the nervous system is the feedback loop. 23 All of these descriptions of feedback in performance were essentially representing a trend toward understanding man as a teleological mechanism: one capable of purposeful action. Miller, Galanter and Pribram (1960) reviewed much of the work in this area and, subsequently, developed one of the f i r s t definitive models for the teleological role of feedback in performance. Tote Mechanism. The mechanism they described is essentially a reaction to the notion that the simple reflex arc forms a basis of behaviour. Sherrington (1906) who examined, in detail, various reflexes, particularly warned that the reflex was too simple a phenomenon to account for behaviour. Even the simple stretch reflex may not be a true reflex at a l l . Indeed, Miller, Galanter and Pribram (1960) have presented physiological evidence which tends to indicate that the central nervous system compares the patterns coming from the muscle spindle with some centrally originating "spindle control" signal pattern inorder to determine what contribution the muscle contraction has made to the "spindle sensing" pattern. The neural mechanisms involved here cannot be explained as a simple reflex arc - a much more complex system of monitoring, or testing, is involved in reflex action. The general pattern of reflex action they described (Miller, Galanter and Pribram, 1960: 26) involves testing "the input energies against some criteria established in the organism, to respond i f the result of the test is to show an inconguity, and to continue to respond until the inconguity vanishes, at which time the reflex is terminated". In this mechanism there is feedback from the result of action to the representation of the intended action. A "recursive loop" is implicated in this process. 24 The basic unit of this mechanism is the TOTE: test-operate-test-exit. The TOTE asserts, simply, that the operations performed by an organism are constantly guided by the outcomes of various tests. The test phase is used both to indicate i f the operation to be performed is appropriate and to determine i f the outcome of the operation meets the expectations. Thus, the TOTE represents the basis for behavioural Plans. The test phase involves the specification of whatever knowledge is necessary for the comparison to be made. The operation phase represents what the organism does as a result of this comparison. The exit phase involves the subject's observation that the inconguity has been eliminated and no further action is required. This mechanism proposed by Miller, Galanter and Pribram (1960) is a highly cognitive one. They see all of behaviour being formed of Plans which are comprised of subplans each of which is comprised of its own TOTE units. The Image represents the value of a certain action which they use to explain which particular Plan will be chosen. The Plan represents the intention, is formed of the TOTE units, and explains how and why a particular action is performed. Both Images and Plans form the motivations of persons which, they see, as being extremely important concepts in explaining behaviour. The TOTE mechanism in many respects represents a reflex action; however, Miller, Galanter and Pribram (1960) place great emphasis on the role of the feedback loop. Throughout an action the subject uses feedback to compare, in the test phase, with some representation or idea of the intended action. Although feedback is stressed there is a difference 25 between this theory and other feedback theories regarding what comprises the intended action to which feedback is compared. Here the intended action represents a cognitive structure of action and is not necessarily represented in the form of a visual or kinesthetic image, for example. Other theories, however, notably those proposed by Greenwald (1970) and Adams (1971) do envisage that the feedback from the intended action forms the image. Afferent Synthesis. The afferent synthesis theory is the teleological approach developed by several Russian psychologists, notably Anokhin (1969) and Sokolov (1969). This approach grew out of an attempt to explain cybernetic theory in terms of neurophysiological mechanisms. Although much of the evidence for this theory came from conditioning, especially research related to the conditioned reflex, the conditioning approach taken here, unlike the previous conditioning theories, does include the aspect of purposefulness in behavioural action discussed by Adams (1971). Anokhin (1969) proposed that the major component in the formation of any adaptive behavioural act involves afferent synthesis. He holds the approach proposed by Pavlov (1932) that the afferent part of the central nervous system is the "creative" one; he ascribes a merely "technical" role to efferent functions. According to Anokhin afferent synthesis invariably precedes decision making and assists in the solution of three fundamental problems in any behavioural act: what to do, how to do i t , and when to do i t . It essentially involves a synthesis of all afferent information enabling the subject to decide among millions of possible movements of which 26 he is capable (response selection) and to determine i f what he has done meets with expectations based on stored afferent information (error detection). The problem of what to do is solved by the dominant motive component of afferent synthesis. The dominant motive is created either (1) as a result of hormonal or metabolic action in the subject (eg. hunger) or (2) as a result of previous behaviour in which, for instance, an error may be perceived to have occurred resulting in the motivation to correct the error. Certain physiological properties outlined by Anokhin (1969: 835-841) suggest that sensory information, which has l i t t l e significance for the dominant motive at a given time, is excluded from integration. Environmental afferentation, a second component of afferent synthesis, essentially determines how to perform the behavioural act. This component creates "a broad integrated system of stimuli in the central nervous system, a kind of neural model of the environment because of its relatively constant nature". (Anokhin, 1969: 840) This component represents an afferent image of various behavioural acts which represent, to the subject, how a particular act should be performed. These afferent images remain subliminal for long period of time and can be transformed to an active state by means of activating afferentation. This activating afferentation is the third component of afferent synthesis which determines when a particular act is performed. Anokhin (1969: 841) writes: 27 A ch a r a c t e r i s t i c feature of the activating stimulus i s that i t precisely times transitions of the integrated neural condition already formed in the brain on the basis of previous motivational and environmental afferentation. While the dominant motivational state determines what the organism must do and environmental afferentation determines how i t must do i t i n a given s i t u a t i o n , the activating stimulus determines the t h i r d important a t t r i b u t e of a behavioural act - when to do that which has been suggested by both the motivational state and the environment. In describing how a behavioural act i s formed Anokhin indicated that sensory integration through afferent synthesis was of primary importance. Through practice an afferent and c o l l a t e r a l efferent image of the correct response i s formed. This i s formed as part of the second component of afferent synthesis. This image i s referred to as the "acceptor of ef f e c t " or the "acceptor of action". When a behavioural act i s performed, feedback from the response i s compared with t h i s image. If a discrepancy i s detected, an analytic orienting response i s i n i t i a t e d which proceeds, through afferent synthesis, to select the correct response, i e . , the efferent c o l l a t e r a l . The entire process of afferent synthesis i s further f a c i l i t a t e d by centrifugal effects which act to reduce the threshold of peripheral sense organs primarily involved i n the afferent synthesis as determined by the dominant motive. According to afferent synthesis theory, error correction and the error detection models are imaged i n one mechanism, the acceptor of ef f e c t . Closed-loop Learning Theory. The most recent t e l e o l o g i c a l theory which bears some resemblance to the previous theories and which i s of primary concern i n this study, i s the closed-loop theory of learning. Closed-loop motor learning theory originated, mainly, from an attempt by Adams to explain how feedback was used i n motor performance (Adams, 1968). 28 However, some of the precise theoretical network of the theory was developed from two different observations in verbal learning studies. First, evidence provided by Luh (1922) and Postman and Rau (in Adams, 1967: 252) showed that there was a differential sensitivity between recognition and recall of verbal items in favour of recognition. Further, studies by Davis, Sutherland and Judd (1961) and Bahrick and Bahrick (1964) showed that recall and recognition could be independently manipulated. Therefore, based on this evidence, i t is possible that each represents a different memory state (Adams, 1967). Recall may represent a memory state based upon the ability to reproduce the item. Recognition, on the other hand, appears to be a state of memory based on the ability to recognize the item when presented among other verbal items. Further, as discriminability affects recognition of a stimulus in the realm of perception and since Bahrick and Bahrick (1964) demonstrated a similar effect of discriminability in verbal recognition, i t is possible that recognition involves a large perceptual component. Thus, recognition may represent a memory state which is based on a perceptual trace. Consequently, based on this f i r s t observation from verbal learning the closed-loop learning theory is proposed to be based on a dual memory state. This represents a significant shift from the single memory state hypothesis espoused by Anokhin (1969). The existence of a dual memory state has been supported by research in verbal learning (Adams and Bray, 1970) and has received some preliminary support in motor learning (Marshall, 1972). A second observation from verbal learning which aided in formulating the closed-loop theory was that learning could progress in the absence of KR. In a study by Eimas and Zeaman (1963) i t was observed that the latency of responding with the correct response occurred in non reinforced trials of 29 a paired associate paradigm. They suggested that the subject may have stored a representation of the correct response pair, and, when required to make a response, he scans his memory and compares his response with the stored traces; i f a match occurs, reinforcement results which leads to a strengthening of the associative bond. Further, they indicated that, i f the subject is scanning his memory, he will respond more quickly on trials in which he can find the matching response than on trials in which he cannot, simply because i t takes longer to continue scanning in the latter. Support regarding learning in the absence of KR is found in work by Jones (1962) who found that the probability of the correct response increased on non reinforced trials. This observation of learning in the absence of KR was important in the formation of the concept of subjective reinforcement. This refers to the idea that the subject can reinforce his own responding by finding a match between intrinsic feedback and the perceptual trace. Some preliminary evidence by Williams and Roy (1972) indicates that subjective reinforcement may be occurring in motor learning. The closed-loop model of learning has as its basic components feedback, error detection and error correction. Feedback is composed of two main types: extrinsic feedback or KR and intrinsic feedback, either interoceptive (eg. proprioception) or exteroceptive (eg. vision, audition). These modes of feedback assume primary importance at different stages in the acquisition of a motor s k i l l . Extrinsic feedback or KR is seen to assume most importance during the early states of skill acquisition when there is a large cognitive component in the task. Adams referred to this period as the verbal-motor stage of learning; i t is similar to the early or cognitive phase of learning proposed by Fitts and Posner (1969). At the beginning of this early stage of acquisition, the subject is not making 30 responses that are subjectively familiar to him as his error is so great. The KR provides the basis for developing a perceptual trace, the internal feedback representation of correct performance, which, later in this stage, aids the subject in recognizing when he has made an error. The intrinsic stimuli fed back from the responding limb are affected by the KR in such a way that they become much less variable since reported error (KR) decreases with increased accuracy of the response. The subject becomes progressively more familiar with the "feeling" of the correct response since the stimuli fed back from the response become more alike, ie., they are all associated with the correct response. Further, Adams (1971) indicated that the perceptual trace is not really a single entity but rather a distribution of traces where the most frequently occurring intrinsic stimuli (mode) come to describe the perceptual trace. As a consequence, i t appears that the perceptual trace becomes stronger due to the impingement of similar stimuli thus enabling modal value of the distribution to be the stimuli associated with the correct response. This strengthening of the perceptual trace, which seems to occur through increased familiarity with the correct stimuli, gradually enables the subject to detect error solely on the basis of feedback stimuli. This internal process of error detection is referred to as subjective reinforcement (Adams, 1971). At this point KR can be removed and performance will not deteriorate. This marks the onset of the second or motor stage in performance and is characterized by the predominance of intrinsic feedback as the source of information for detecting error. Other basic components of the model are error detection and error correction. After the perceptual trace is well established, the response-produced stimuli are compared with this trace. Any deviation of the current feedback from the reference mechanism is detected as error. Thus, there is 31 an attempt made to reduce or correct this error by firing a correcting response. It can be seen that in this model error is motivating, i.e., i t prompts the subject to correct the error. In the early stages of learning KR provides the motivation, while in the later motor stage error provides the controlling motivation (Adams, 1971). So far i t appears that the learning and performance are primarily under the control of the perceptual trace, i.e., i t provides the reference mechanism which is used to detect error. The major question in the theory comes at this point. Can the mechanism which detects error also be used to initiate the correcting response? Several researchers, notably the Russian psychologist, Anokhin, favour the one trace or one memory state hypothesis. Adams (1971: 121), however, indicates a basic criticism of this approach in the following: The flaw in this approach is a failure to account for error detection...The agent that fires the response and the model that tests i t must be different because without a difference we would not know that an error has occurred. Further evidence is that the use of the perceptual trace depends on feedback which occurs after the response begins. Some other mechanism is needed to initiate the response in the fi r s t place. Finally, recall is response production while recognition is identification of a response. As discussed previously, recall and recognition can be independently manipulated. Based upon these arguments, Adams proposed that the firing of a response is based upon another memory state involving a memory trace. The 32 memory trace is considered to be independent of feedback. That is, unlike the perceptual trace which requires feedback in order to be used, the memory trace can be fired without reference to feedback. As such i t appears similar to a motor program. The role of the memory trace is "to select and initiate the response, preceding the use of the perceptual trace". (Adams, 1971: 125) The memory trace must be "cued to action"; i.e., the subject must be required to recall the response. The memory trace strength develops as a function of practice trials and KR. In an attempt to further delineate the closed-loop model, Adams, Goetz and Marshall (1972) investigated the effect of differential amounts of practice (15 or 150 trials) on the development of the perceptual trace and the effect of changing the characteristics of feedback, during KR withdrawal, from those which prevailed during acquisition in a simple arm movement task. Three variables were manipulated in this study: level of learning (high or low), feedback in acquisition (augmented or minimal), feedback in KR withdrawal (augmented or minimal). Eight conditions were involved in the study: LAA, LAM, LMA, LMM, HAA, HAM, HMA and HMM. The fi r s t letter refers to the amount of practice with KR where L is low (15 trials) and H is high (150 trials). The second and third letters refer to the type of feedback in acquisition and in KR withdrawal, respectively. Augmented feedback (vision, audition and augmented proprioception) is indicated by the letter A while M indicates minimal feedback where visual, auditory and proprioceptive feedback were reduced. Thus, condition HMA, for instance, would involve 150 trials in practice with minimal feedback and during KR withdrawal augmented feedback was given. Fifty trials were given in KR withdrawal. There were 20 subjects in each group. 33 The authors found that i f feedback was altered from that which was present in acquisition performance deteriorated significantly. Further, they found that giving subjects 150 trials of practice as opposed to 15 trials resulted in significantly better performance (absolute error) on non KR trials only when the type of feedback in these trials was the same as that in acquisition. When feedback was changed on these trials from that present in acquisition, the conditions with 150 trials experienced as great error as those which received only 15 trials. These results indicated, f i r s t , that the perceptual trace developed in acquisition was specific to the type of feedback present in acquisition. Secondly, the strength of the perceptual trace was augmented by increased practice. This finding had the qualifi-cation, however, that increased practice only aided performance when KR was withdrawn i f the feedback remained the same as that during acquisition. If feedback was changed, the subject was unable to use the perceptual trace to detect error, regardless of the amount of practice experienced to develop i t . Other evidence by Adams, Marshall and Goetz (1972), Marshall (1972), Schmidt and White (1972) and Williams and Roy (1972) is consistent with a closed-loop interpretation for the control of motor performance. In summarizing this section on feedback in performance i t is evident that all these theoretical approaches have the common requirement for intrinsic feedback. The conditioning theories suggest that, through contiguity, intrinsic feedback comes to replace external cues as a stimulus for responding. The teleological theories, on the other hand, regard man as an information processor in which feedback from a response is compared 34 to some central feedback representation in order to detect error. This error information is subsequently used to select some correcting response. Preprogramming in Motor Performance The preprogramming approach to the control of motor performance is characterized by its emphasis on central, rather than peripheral (feedback), information. Much evidence has accumulated in favour of this approach. The major sources of this evidence will be presented in this section beginning with a brief description of the contributions of Karl Lashley. Other evidence provided by research on the reduction of feedback, on • efference in perception and on tracking and anticipation will be discussed. Finally, theoretical approaches to explain the mechanisms involved are presented. Lashley - On Central Control of Performance One of the earliest proponents for the central organization of behaviour was Lashley (1917, 1951). During his period of research the most popular theory of feedback in performance was response chaining. Thus, most of his research was an attack on this theory. The response chaining hypothesis emphasized the role of proprioception. Lashley and his colleagues (Lashley and McCarthy, 1926; Lashley and Ball, 1929) directly manipulated i t in rats by either severing the afferent nerves carrying proprioceptive information to the brain or by lesioning the cerebellum, the coordination center for movement. Subsequent maze learning was successful, although motor coordination was poor. Pre- and post-operative movements 35 bore l i t t l e resemblance to one another, a finding which the response chaining hypothesis would not predict. Thus, Lashley concluded that the maze learning was centrally organized without proprioception (Lashley, 1951). He also illustrated the central control of performance with the example of a musician whose finger strokes may reach 16 per second - too fast for proprioceptive feedback to operate (Lashley, 1951: 123). Other examples were drawn from language to demonstrate that words of a sentence are under central rather than peripheral control. Although accurate performance in the absence of proprioception does reduce the plausibility of the response chaining hypothesis, evidence for preprogramming must show that learning can occur with all feedback loops eliminated. Such evidence is provided in several research papers dealing with reduction of feedback. Two primary means of reducing feedback are surgical excision of nerve trunks carrying afferent information or some form of non surgical method such as nerve compression block (Laszlo, 1967a). The fi r s t areas of research to be discussed are those dealing with surgical reduction of feedback. Reduction of Feedback Surgical Methods. Lashley (1917) investigated the movement accuracy of a man who, as a result of a war injury, was left with virtually no pro-prioceptive or tactile sensitivity in his left leg. The man was not able to perceive or reproduce passive movements of the leg, but was able to reproduce, as accurately as the normal leg, movements that had been previously produced by him without visual information. Further, on several 36 occasions, the subject indicated that he had made a movement longer than he intended; indeed, these movements were longer than others intended to be the same length. This evidence would seem to imply that the learning was centrally organized, and further, that there may have been a central feedback loop in which the issued command is compared with that intended. Research carried out by Wilson and his colleagues (Wilson, 1961, 1964, 1968; Wilson and Gettrup, 1963; Wilson and Wyman, 1965) demonstrated that removal of all of the proprioceptors of the locust associated with wing movement during flight did not disrupt appreciably the coordinated movement of the wings, although the frequency of movement was reduced by one-half. Further, Wilson and Wyman (1965) varied the timing of electrical stimulation of the wing proprioceptors. They found that the phasing of the wing movement remained unchanged. Therefore, i t appeared that timing of the output came from some central neural organization, not the feedback input. These findings led Wilson (1968) to conclude that there was a central programming of wing movement. Taub and his associates (Taub and Berman, 1963; Taub and Berman, 1968; Taub, Bacon and Berman, 1965) performed deafferentation on monkeys and found that they were able to relearn movements - in the total absence of peripheral feedback information. Support for this deafferentation research is found in other research carried out by Beck and Doty (1957) and Solomon and Turner (1962) which indicated that classical conditioned responses can be learned, and that extinction and performance of an instrumental conditioned response can occur while animals are effectively paralyzed. These findings led Taub, Bacon and Berman (1965: 278) to comment regarding the mechanism involved: 37 One possible mechanism requires the existence of a purely central feedback system that could, in effect, return information concerning future movements to the CNS before the impulses that will produce these movements have reached the periphery. Just such a mechanism as described by Taub is suggested by physiological information provided by Chang (1955), Li (1958) and Kuypers (1960). Non Surgical Methods. The second technique for reduction of feedback involves a non surgical method, such as a nerve compression block. Chase (Chase, Harvey, Standfast, Rapin and Sutton, 1961; Chase, Rapin, Gilden, Sutton and Guilfoyle, 1961; Chase, Sutton and Rapin, 1961) investigated the effects of decreasing and delaying feedback on keytapping performance. In one representative study Chase, Harvey, Standfast, Rapin and Sutton (1961) decreased vision, audition, proprioception and tactile stimuli. Vision and tactile stimuli were prevented by means of a blind fold and an application of xylocaine to the finger tip, respectively. Proprioception and audition were not actually prevented, rather, the stimuli were masked. Proprioception was masked using a vibrator on the arm while audition was masked with white noise. In another part of the same study, auditory, visual and tactile stimulation were delayed. The results indicated that both decreasing and delaying feedback disrupted performance. In examining the reduction of feedback the single most detrimental reduction in feedback occurred when proprioception was reduced in that as much disruption in performance, measured by changes in frequency and amplitude of tapping, occurred in this condition as in a condition in which all feedback channels were blocked. 38 Delaying feedback also resulted in a disruption of performance. In explaining how delaying of feedback affected performance Chase, Rapin, Gilden, Sutton and Guilfoyle (1961) proposed that the feedback stimuli were returning in an improper sequence. They hypothesized that the central nervous system used the delayed feedback to detect error. Consequently, the system would initiate a corrective response based on this delayed feed-back resulting in severe disturbance in performance since these error corrections lagged behind the actual performance. These findings indicate that sensory feedback is important in performance and Chase went on to develop a model of performance based on the use of sensory feedback (Chase, 1965a, 1965b). Provins (1958) investigated the effect of rate decrement in finger oscillation under xyolcaine nerve block in presence of reduced vision and audition. He found virtually no decrement in oscillation rate thus seemingly refuting Chase's view that feedback was important. Laszlo (1967a, 1967b) found that, like Chase, the loss of kines-thetic feedback was more detrimental than the loss of vision or audition. However, she criticized both Chase and Provins regarding their techniques. According to Laszlo (1967a) Chase's technique of applying a vibrator was not really preventing feedback, rather, i t was only masking i t . Thus, possibly the decrement in performance observed by Chase was not due to the reduction of feedback, but rather to the masking itself. She criticized Provins (1958) since the xylocaine application he used would have only prevented tactile stimuli. The kinesthetic stimuli were left completely in tact. Finally, she criticized both researchers for not attempting to 39 train subjects under reduction of feedback. Research by Taub (eg. Taub and Berman, 1963) showed that primates could learn when no peripheral feedback was present. Therefore, humans should be able to do the same. In initiating her research Laszlo (1966) developed a technique called a nerve compression block that, unlike the methods employed by Provins (1958) and Chase (eg. Chase, Rapin, Gilden, Sutton and Guilfoyle, 1961), actually did prevent both tactile and kinesthetic feedback. Using this technique Laszlo (Laszlo, 1967a, 1967b, 1968; Laszlo, Shamoon and Sanson-Fischer, 1969; Laszlo and Manning, 1970) carried out a series of experiments demonstrating that, indeed, subjects could learn in the absence of feedback. Efference in Perception Other research demonstrating that preprogramming is important in motor performance comes from research on the role of efference in perception. Paillard and Brouchon (1968) found that active arm movement was more accurate in reproducing a movement length than passive movement. In terms of afferent feedback both active and passive movements provide approximately the same cues. In active movement, however, an efferent control signal is involved while in passive movement no such signal is present. Paillard and Brouchon (1968) attribute the increased accuracy in the active condition to the presence of this "motor outflow". When the efferent signal is elicited i t travels to the musculature and is also believed to have a "corollary discharge" which is sent to some central storage mechanism in the CNS (Sperry, 1950; Held, 1961). The term "corollary discharge" essentially describes what Hoist (1954) designated as an "efference copy" or what 40 Merton (1964a) described as the "sense of effort". Hoist (1954) attributes veridical perception to the presence of this efferent discharge. Intrinsic feedback which is dependent on the subject's own movement, ie., in the presence of efference, is termed reafference by Hoist (1954), while feed-back not dependent on the subject's own movement (extrinsic feedback) is called exafference. Veridical perception occurs only when reafference is matched with its corresponding efferent command. Further, Held (1961) attributes perceptual adaptation to visual rearrangement, caused by dis-placing prisms, to the presence of an efferent command. Finally, the presence of such an efference mechanism appears to account for the ability of deafferented monkeys to perform accurate reaching even without visual control (Taub and Berman, 1968). Other evidence as to the importance of efference in perception is found in work on the control of eye movements. Although proprioceptors are present in the ocular muscles, i t appears they are not functional at a conscious level (Merton, 1964b). Adams (1968) suggests that the absence of conscious sensation does not imply absence of function. Nevertheless, much research shows that the primary mechanism for orienting the eyes is provided by efference (Irvine and Ludvigh, 1936; Festinger and Canon, 1965). Indeed, Festinger, Ono, Clarke and Bamber (1967) demonstrate that visual perception is determined by efferent readiness activated by afferent visual input. They indicate that the subject learns the appropriate efferent instructions to be issued from the CNS to direct the eye to move in or*der to bring any particular stimulus on to the fovea. After a great deal of learning has occurred, these efferent instructions are viewed as becoming preprogrammed and brought into a state of immediate readiness for use. 41 Which particular sets of preprogrammed instructions are used is determined by "brightness or color differentials stimulating the retina". (Festinger, Ono, Clarke and Bamber, 1967: 12) Other Evidence for Preprogramming Research on tracking by Higgins and Angel (1970) and work on choice reaction time by Rabbitt (1966, 1967) indicate that errors can be corrected in less time than could be explained on the basis of visual (Keele and Posner, 1968) or proprioceptive (Chernikoff and Taylor, 1952) processing time. Other tracking research by Angel and Garland (1972) also implicates preprogramming. Further evidence in favour of preprogramming comes from research on timing by Schmidt (Schmidt, 1969; Schmidt, 1972; Schmidt and Russell, 1972). In a coincidence timing task, where a subject must coincide a movement with some external event in a certain finite time period, Schmidt (1969; 1972) hypothesized that preprogramming of such a movement would be indicated i f there was a high relationship between the starting time (ST), defined generally as the time between when the subject begins his response and when the external "target" reaches the coincidence point, and the error (algebraic) of performance. If a movement is preprogrammed such that feed-back during the response has no effect, then the time of finishing the response, ie., lapsed time or algebraic error should be highly correlated with the time i t started. A feedback controlled response, on the other hand, would allow feedback to adjust the response during its execution resulting in a low relationship between ST and algebraic error. Consequently, 42 Schmidt (1972) used this correlation measure, converted to a standard score, as the index of preprogramming (IP). In one investigation Schmidt (1969) found that ST was correlated with algebraic error (r=.73) and movement time (MT) (r=.63). MT was not linearly related to algebraic error (r=.04). These findings led Schmidt to suggest that the subject had preprogrammed his response. Since ST was correlated with timing accuracy and with MT, i t appeared that the subject may have been choosing an ST in advance. It appeared that the subject attempted to hold his MT relatively constant while varying his ST in response to reported error on the previous t r i a l . This hypothesis was supported in that the intravariance of ST (.043 seconds) was greater than that of MT (.028 seconds). Further, higher relationships were found for maximal than for moderate speed responses between ST and algebraic error and between ST and MT indicating, possibly, that a greater preprogramming occurred as the execution speed of the response increased. Further study of this relationship between MT, movement speed and preprogramming was carried out by Schmidt and Russell (1972). In this study, two MTs (.150 and .750 seconds) and two movement distances (22.8 and 49.5 cm.) were used in a two by two factorial design. In this design Schmidt was able to directly control both MT and movement speed, thus enabling a close examination of which variable precipated preprogramming. The data were analyzed using the IP mentioned above. The analysis revealed that, reducing the MT from .750 seconds to .150 seconds significantly increased the IP from Z'=.66 to 1.39, indicating that the short MT prevented the use of feedback and required the response to be 43 programmed to a greater extent. Increasing the movement speed, however, by holding the MT constant and varying the movement distance, only slightly increased the IP. These findings strongly implicate the MT as the major determiner of preprogramming. Schmidt and Russell (1972) indicated that the .150 seconds MT response consisted primarily of a preprogrammed segment, while the .750 seconds MT response involved a preprogrammed initial segment plus a feedback-based correction. Keele (1968) has suggested three reasons why preprogramming may be advantageous to performance. First, the degree of attention may be reduced. Some support for this has been found in studies by Noble, Trumbo and Fowler (1967) and Posner (1969). Second, reaction time lag in coinciding with external stimuli may be reduced as successive stimuli may be anticipated. Evidence on motor programming in anticipation is found in Schmidt (1969) and Schmidt and Russell (1972). Finally, preprogrammed movements may be made much faster than those dependent on feedback. Control Mechanisms in Preprogramming Preprogramming theory offers an explanation for the learning of a motor response and for its control following the withdrawal of KR. In acquisition KR is used to develop a motor program while KR and intrinsic feedback develop a standard which may be somewhat analagous to the perceptual trace in closed-loop theory. Following KR withdrawal there are two hypotheses as to how the mechanisms, developed in acquisition, control performance. Keele (1968) has suggested that control is solely under the 44 direction of the motor program which he sees as a sequence of stored commands that is "structured before the movement begins and allows the entire sequence to be carried out uninfluenced by peripheral feedback". (Keele, 1968: 387) Alternately, Laszlo and Manning (1970) have proposed that one of two mechanisms may be involved. One involves purely the development of a motor program as suggested by Keele (1968). The other involves a type of central loop in which the command from the motor program, or motor programming unit as referred to by Laszlo and Manning (1970), is compared to a type of standard in order to detect error, possibly even before the response is initiated. As mentioned above, Taub (in Taub, Bacon and Berman, 1965) favours the latter mechanism stating that a type of central loop system may be involved where information concerning future movements is returned to the CNS before the impulses that will produce them have reached the periphery. Indeed, physiological evidence provided by Chang (1955) and Li (1958) suggest such a mechanism. In summarizing this section on preprogramming theory, i t appears that much evidence has accumulated to strongly suggest such a mechanism. Deafferentation research on humans (Lashley, 1917), monkeys (Taub and Berman, 1968) and locusts (Wilson, 1961) shows that accurate and coordinated performance can occur in the total absence of feedback. Research on the reduction of feedback (Laszlo, Shamoon, Sanson-Fischer, 1969), on the role of efference in perception (Festinger, Ono, Clarke and Bamber, 1967), on tracking behaviour (Higgins and Angel, 1970; Angel and Garland, 1972) on choice reaction time (Rabbitt, 1966, 1967), and on timing (Schmidt and Russell, 1972) implicates the role of preprogramming in performance. In essence this approach to the control of motor performance sees no role for 45 feedback following the withdrawal of KR. Rather, the response is pre-programmed and executed without reference to intrinsic feedback. Discussion In general, the use of intrinsic feedback, as proposed by the feedback theories, or the use of central, efferent information, as suggested by the preprogramming approach, provide a plausible explanation for the control of performance following KR withdrawal. In this study, however, particular emphasis is placed on the closed-loop theory proposed by Adams (1971) and the preprogramming theory proposed by Laszlo and Manning (1970). In comparing these two theories in their explanations for the control of motor performance during KR withdrawal i t was observed that both possess the requirement for some type of control loop. The major difference between these theories, however, regards the necessity for intrinsic feed-back. Only closed-loop theory proposes that intrinsic feedback is necessary. Therefore, i t may be that the type of theory used to explain performance control may depend upon the availability of feedback during the execution of the response. Two types of responses in which there appears to be a difference in the availability of feedback are a self-paced and a ballistic response. A self-paced response is one which is primarily guided by intrinsic feedback, while a ballistic response is one which is executed too quickly for feedback guidance. Further, i t appears that a self-paced response is characterized by a long movement time (MT), while a ballistic response has a relatively short one. In regards to MT, Schmidt and Russell (1972) have shown that the MT of a response is a major determiner of the amount of preprogramming developed. As a consequence, i t may be that a 46 response which has a relatively long MT (>1 second), thus allowing the use of intrinsic feedback during execution (e.g. self-paced), may be ideally suited for closed-loop control in that error detection can occur through a comparison of feedback to the perceptual trace while the response is being executed. On the other hand, a response which has a short MT (<.150 seconds), thus preventing the use of intrinsic feedback during execution, may be more amenable to preprogramming control. This initial distinction is question-able, however, since, through learning, there is no reason why a self-paced task could not also come under the control of a motor program. In addition, Schmidt and White (1972) have argued that a ballistic response may be explained from a closed-loop viewpoint, since the feedback from a ballistic response could be compared to the perceptual trace of that response after the movement was completed. In fact, they interpret their results as supporting a closed-loop control mechanism in a ballistic response since the subjects were able to learn on non-KR trials, and there was a relation-ship between subjective and objective error. Both these findings are in the direction predicted by closed-loop theory in terms of the use of the perceptual trace and subjective reinforcement, respectively. Nevertheless, i t is suggested here that the results could also be interpreted in terms of motor programming theory as this theory would predict similar results. Evidence by Taub and Berman (1968), Laszlo (1967) and Laszlo, Shamoon and Sanson-Fischer (1969) shows that learning can occur when all feedback is eliminated. In addition, Lashley (1917) found that his deafferented subject could perceive whether he had moved his leg past the criterion distance based, i t seems, purely on the efferent command. In order to establish which control mechanism is operating i t is necessary to consider the proposed role of feedback in each type of response. 47 The basic assumption here, which leads to an hypothesis implicating differential control mechanisms, is that i f preprogramming has occurred in a given response during acquisition, then changing the characteristics of the feedback or preventing the use of feedback, following KR withdrawal, should not result in performance decrement, since peripheral feedback is not used in performance entirely under the control of a motor program. On the other hand, manipulation of feedback by either changing the characteristics of feedback during KR withdrawal from those which prevailed during acquisi-tion, (Adams, Goetz and Marshall, 1972), or by reducing feedback (Chase, Sutton and Rapin, 1961) have been shown to have a detrimental effect on a response under closed-loop control. Thus, two experimental methods are proposed here to determine the mechanism in each type of response. One involves the reduction of available feedback during the post-acquisition or KR withdrawal stage of learning; the second involves preventing the subject from using feedback by occupying his limited processing capacity system through performance of a secondary task. This technique has been used by Posner and Boies (1971). It is proposed that through these two methods an adequate differentiation between closed-loop and motor programming mechanisms can be made for both self-paced and ballistic responses. Reduction of peripheral feedback by reducing sensory input or by having the subject's processing capacity engaged by a secondary task while he is performing a criterion task should inhibit performance i f a closed-loop mechanism was developed during the acquisition phase. These experimental manipulations inhibit the subject from matching the response-produced feedback with the perceptual trace thus denying him use of the mechanism that Adams (1971) postulates controls 48 movement execut ion. On the other hand, i f a motor program was developed these manipulations should have no e f fec t on performance, s ince per ipheral feedback i s unnecessary fo r execut ion. CHAPTER III METHODS AND PROCEDURES Subjects Sixty students from undergraduate and graduate studies at the University of British Columbia were used as subjects. There were 36 male students and 24 female students involved. Three male students and two female students were assigned to each of the twelve groups in a systematic, unbiased fashion thus making a total of five subjects per group. Only right handed students were used as subjects. Apparatus The apparatus used in this experiment was a modification of one originally designed by Ellis (1969). A pictorial description of the apparatus is shown in Figure 1, Appendix A. Track. The track rested on a table measuring 48.0 inches by 48.0 inches by 36.0 inches. The top of the table was made of 1.0 inch plywood and was bolted to the frame. The legs and frame were made of angle iron which was bolted together. The track was situated about 5 inches from one edge of the table. This edge was the one in front of which the subject sat. The base of the track was made of stainless steel and was 34.0 inches long, 3.0 inches wide and 0.50 inches thick. At each end of the 49 50 base was an-end plate which was 7.0 inches high, 2.75 inches wide and 0.75 inches thick. The plate at the one end was adjustable. Into these plates a U-shaped cut, 3.25 inches deep and 0.75 inches wide, was cut. At the top of each plate, set in the U groove, was a brass pulley. The pulley at the end of the track where the flywheel was situated measured 3.50 inches in diameter and 0.50 inches in width, while the pulley at the other end was 3.50 inches in diameter and 0.25 inches in width. A shallow groove was ground around the outer perimeter of these pulleys. A braided wire, 0.125 inches in diameter, was wrapped around these pulleys and sat in the grooves. This wire formed a long loop from one end of the track to the other and was attached to a steel cursor. In each end plate, approximately 0.50 inches below the base of the U-groove was bored a 0.60 inch hole into which fitted the end of a cylindri-cal bar 0.50 inches in diameter. This bar was secured in the end plate at the flywheel end of the track by set screws. At the other end the bar was threaded. Brass washers 2.0 inches in diameter by 0.50 inches thick were screwed onto the bar on either side of the end plate. By adjusting these set screws the position of the end plate could be adjusted. This adjustment permitted the tautness of the control wire to be adjusted. Cursor. A cursor reamed from stainless steel slid on the bar described above. The cursor was shaped like an inverted T and was, basically, formed of two cylinders which were fused together into the T shape. The lower horizontal part of the cursor measured 2.50 inches long with an outside diameter of 1.0 inches and an inside diameter of 0.75 inches. This was the part which slid on the bar. Projecting 90 degrees vertically to the horizontal cylinder was a vertical cylinder 1.75 inches 51 high and 1.0 inches in diameter. At the top was a cap which could be loosened to enable the ends of the control wire to be slid under. The cap was then tightened with set screws thereby securing the control wire to the cursor. On one side of the cursor projecting 90 degrees to the horizontal cylinder was a cylindrical handle 4.0 inches long and 0.5 inches in diameter. Projecting at 90 degrees to the other side of the cylinder was a 2.0 inch round head screw 0.25 inches in diameter. This screw was used to trip the microswitches. Flywheels. A flywheel was placed on one end of the track on a 0.50 inch threaded steel bar which projected through the side of the end plate to a distance of 2.50 inches. The other end of this bar was attached to the brass pulley. A rectangular hole, 14.0 inches by 2.5 inches was cut in the table to accommodate the flywheel. The flywheel added tension to the cursor movement by increasing the force required to move i t . This tension was created because the cursor was attached to the control wire which wrapped around the pulleys. When the cursor was moved, i t moved the control wire which, in turn, moved the pulleys. Since the pulley was attached to the flywheel, the movement of the cursor moved the weighted flywheel. Two flywheels were used: a heavy and a light one. The heavier one was used in the high feedback condition while the lighter one was used in the low feedback condition. Both flywheels were 12.0 inches in diameter. The heavier one was cut from stainless steel, was 0.25 inches thick and 52 weighed 32 ounces. The lighter wheel was made from galvanized iron, was 0.125 inches thick and weighed approximately 12 ounces. Microswitches. Along one side of the track was a row of 8 micro-switches (Cutler-Hammer Model V3L-6-D8). The distance between the contact point of the first microswitch and the last microswitch was 27.5 inches. The other distances for the microswitches are shown in Figure 2, Appendix A. The first and last microswitches were connected through a latch relay to a Hunter Klockounter timer Model 120A. The time interval between these micro-switches gave the overall time taken to move the cursor the distance of the track and was the dependent measure. The six microswitches between the two end switches were used to measure the intra-track times. Each pair was 4.5 inches apart. The switches were paired together to give three intra-track times. Each of the three pairs was connected to one of three timers in a Monsanto Timer Model 101B. The f i r s t microswitch in each pair started the clock the pair was connected to, while the second switch stopped the clock. The interval thus timed was printed on paper tape for all three timers by means of a Digitec Digital Printer Model 691. In order to assure that the screw on the cursor stayed at the same level as the cursor traversed the track so that i t would hit each micro-switch the handle which projected from the other side of the cursor passed along a groove cut out of a piece of fibre board. This groove was 0.50 inches wide and 34.0 inches long and was cut into a wall of fibre board 38.0 inches long and 4.0 inches high. This board was attached at either end to wooden blocks which were secured to the table near the end plates of the track. 53 S e c o n d a r y T a s k . A b o v e t h e t a b l e on w h i c h t h e t r a c k was s i t u a t e d was a n o t h e r , s m a l l e r t a b l e 18.0 i n c h e s h i g h , 40.0 i n c h e s l o n g a n d 24.0 i n c h e s w i d e . T h e t o p o f t h e t a b l e was made o f 1.0 i n c h p l y w o o d w h i l e t h e f r a m e a n d l e g s w e r e made o f a n g l e i r o n . On t h e l e n g t h o f t h e t a b l e o p p o s i t e t h e s u b j e c t a n d o n e i t h e r s i d e o f t h e t a b l e was a 6.0 i n c h w a l l o f f i b r e b o a r d . I n t h e w a l l o f f i b r e b o a r d d i r e c t l y f a c i n g t h e s u b j e c t w e r e p l a c e d s i x 1.5 v o l t f l a s h l i g h t b u l b s , s u c h t h a t t h e r e w e r e two p a r a l l e l r o w s o f t h r e e b u l b s . T h e b u l b s i n e a c h r o w w e r e 4.0 i n c h e s a p a r t w h i l e t h e two r o w s w e r e 4.0 i n c h e s a p a r t . T h e t o p b u l b i n e a c h r o w was 1.0 i n c h e s f r o m t h e t o p o f t h e f i b r e b o a r d , w h i l e t h e b o t t o m b u l b i n e a c h row was 1.0 i n c h e s f r o m t h e t o p o f t h e t a b l e . T h e b a c k g r o u n d a r o u n d t h e b u l b s was p a i n t e d o r a n g e a n d a b l a c k number was p r i n t e d u n d e r e a c h b u l b . T h e b u l b s w e r e n u m b e r e d c o n s e c u t i v e l y g o i n g f r o m r i g h t t o l e f t . T h e b u l b s w e r e c o n n e c t e d i n p a r a l l e l t o a 6 v o l t b a t t e r y . E a c h b u l b was a l s o c o n n e c t e d t o a s e p a r a t e m i c r o s w i t c h ( H o n e y w e l l M o d e l 8N101) w h i c h w e r e c o n n e c t e d i n p a r a l l e l t o t h e b a t t e r y . T h e m i c r o s w i t c h e s w e r e n u m b e r e d s u c h t h a t t h e y c o r r e s p o n d e d t o t h e n u m b e r e d l i g h t w h i c h t h e y a c t i v a t e d . On t h e f r o n t e d g e o f t h i s s m a l l t a b l e d i r e c t l y o p p o s i t e t h e l i g h t s o n t h e b a c k e d g e was a s t y r o f o a m c h i n r e s t on w h i c h t h e s u b j e c t p l a c e d h i s c h i n d u r i n g t h e e x p e r i m e n t . E x p e r i m e n t a l D e s i g n T h e e x p e r i m e n t was d i v i d e d i n t o two p h a s e s . P h a s e o n e was t h e l e a r n i n g p h a s e w h i l e p h a s e two was KR w i t h d r a w a l o r p o s t a c q u i s i t i o n p h a s e . 54 In the f i r s t phase two learning conditions, high and low, and two response types, ballistic and self-paced, were defined. All subjects learned under the high feedback (HF) condition. In the high learning condition a 2 x 150 (response type x trials) factorial design with repeated measures on the second factor was involved. In the low learning condition a 2 x 15 (response type x trials) factorial design with repeated measures on the last factor was used. Thus, in phase one four groups were involved: high learning-ballistic (HL-B), high learning-self-paced (HL-SP), low learning-ballistic (LL-B), and low learning-self-paced (LL-SP). Phase two was the KR withdrawal phase and each of the four groups in phase one was divided into three feedback conditions: high feedback (HF), low feedback (LF) and no feedback (NF). The design was a 2x2x3x50 (type of response x level of learning x level of feedback x trials) factorial design with repeated measures on the last factor. In this phase, twelve groups were involved having fifty trials each. Procedures General Procedures. The subject was seated in front of the apparatus, with his chin on the rest, looking at the display panel. In this condition he was not able to see his hand, the apparatus or any distracting stimuli as a 6.0 inch wall of fibre board extended around the sides and the back of the upper platform of the apparatus on which his chin rested. The subject was instructed to remain in this position during the execution of the timing response. However, he was allowed to remove his chin from the rest and relax during the inter-trial interval (ITI) of 10 55 seconds. All visual cues regarding arm movement or the movement of the cursor were prevented although the subject could view the apparatus during the ITI. The beginning of each trial was signalled by the sound of the buzzer. The subject grasped the lever with two fingers resting on top and two at the side. He was instructed to begin his response as soon as possible after the buzzer sounded. Subjects using the SP response moved the cursor the entire length of the track while those using the B response moved the cursor only 4.0 inches after which they released i t . In both responses the subject's task was to propel the cursor so that i t traversed the distance of the track in one second. The subject returned the lever with his left hand. Five pre-experimental trials without KR were given to each subject. During these trials the subject used the type of response which he was to use during the experiment. He was told to simply feel how the cursor moved along the track. Phase One. Phase one lasted about 30 minutes for subjects in the HL group and approximately 3 minutes for the LL group. Directional quantitative KR was given to the subject in .01 second units. The experimentor, however, recorded in .001 second units. Phase Two. This phase involved performing the timing response in the absence of KR. Also, each condition in phase one was separated into three feedback conditions: high, low or no feedback. Fifty trials were given to each subject. Subjects in the HF condition performed on the 56 timing task as in phase one. Subjects in the LF condition performed only the timing task but a lighter flywheel was attached. Subjects in the NF condition, however, were required to perform a secondary task as well as the timing task. In this condition as soon as the subject began his timing response, the experimentor began to flash on the lights on the display panel. These six lights were flashed on in a random order. The random order for each of the 50 trials had been determined prior to the experiment and was recorded on a sheet of paper. The same random orders were used for each subject. One light was flashed on approximately every .250 seconds until all six had appeared. Each light appeared only once. The subject was told to watch these lights while he attempted to maintain a one second timing response. Before returning the cursor during the ITI the subject was required to recall overtly the sequence these lights went on in. The experimentor recorded any errors made by the subject in recalling this sequence by comparing the subject's response to the random order sequence which he had presented. Finally, the time taken to traverse each of three 4.5 inch intra-track segments was recorded for all subjects using the SP response. Experimental Conditions Self-Paced Response. The subject was permitted to move the cursor across the entire length of the track. During the traversal of the track he could accelerate or decelerate at will although he was told to attempt to make the movement of his arm as constant as possible in terms of velocity. 57 Ballistic Response. In this condition the subject was only allowed to move the cursor 4.0 inches along the track after which he had to release i t . The release point was indicated to the subject by means of a small rubber flange raised 8.0 inches above the surface of the table at the release point. His wrist passing over and touching this flange signalled him to release the cursor. The subject was not permitted to view either his hand movement or the movement of the cursor. The distance to the release point in this condition was determined after consideration of the velocity of the cursor over the entire track and the movement time (MT) of the response to the release point. The fi r s t consideration was the cursor velocity. The average velocity of the cursor moving over the 27.5 inch distance in 1 second would be 27.5 inches per second. An accurate timing response would be associated with this average cursor velocity. The other consideration was that of the MT of the response. Evidence presented by Schmidt and Russell (1972) indicated that a response with a MT of .150 seconds or less necessitated that i t be preprogrammed. Thus, an attempt was made to assure that the MT of the ballistic response was about .150 seconds. Since the average velocity of an accurate response was 27.5 inches per second and since the MT of the ballistic response should be about .150 seconds, the distance of the ballistic response was determined by simple manipulation of the equation, D=VT, where D = distance, V = velocity, and T = time. Therefore: D = 27.5 inches/second x .150 seconds = 4.125 - 4.0 inches 58 Learning Conditions. Two learning conditions were used: high and low. The high learning (HL) condition involved the subject performing 150 trials of practice with KR. In the low learning (LL) condition the subject only had 15 trials of practice with KR. KR Withdrawal. This phase was the second one and involved the performance of the timing response with no external KR. In addition, each of the 4 groups in phase one was divided into three feedback conditions: high feedback (HF), low feedback (LF), no feedback (NF). Finally, the time to traverse each of the three 4.5 inch intra-track segments was recorded for each subject using the SP response. These intra-track times were used to determine i f the subject was altering his response once i t had begun. Closed-loop theory would predict that the subject would alter his response during its execution as he is continually comparing feedback to the perceptual trace in an attempt to nullify any error in his response. Pre-programming theory, however, would suggest that the subject would not alter his response during its execution because feedgack is not used. Therefore, i f there are large differences among these times a closed-loop mechanism would be suggested. Small differences might suggest a preprogramming mechanism. Feedback Conditions. The high feedback (HF) condition involved moving the cursor against a resistance created by the heavier, weighted flywheel. The low feedback (LF) condition involved moving the cursor against a resistance created by the lighter flywheel. Also in this LF condition earphones were placed over the subject's ears to reduce the auditory feedback from the movement of the cursor. 59 The no feedback (NF) condition involved utilizing the subject's channel capacity with a secondary task thus preventing him from using the intrinsic feedback provided by moving the cursor. The secondary task involved recalling the sequence of six lights which were flashed on during the execution of the primary task. Immediately after the beginning of the primary timing response, the experimentor began flashing the lights on the display panel. During the ITI the subject was required to recall the sequence that these lights went on in. Thus, the subject's attention was occupied both during the execution of the primary timimg response and during the ITI. Recent research made i t seem important to utilize the subject's attention during both these periods. Posner and Keele (1970), investigating verbal rehearsal, showed that., although the encoding of a stimulus did not require channel capacity, any further mental operations such as rehearsal did, while work on closed-loop theory related to ballistic response (MT<.150 sec.) by Adams (1971) and Schmidt and White (1972) suggested that, although the subject could not use feedback during the execution of the response, he was able to use the feedback from the response during the ITI in order to develop a perceptual trace. The findings of Posner and Keele (1970) suggested that a subject may be able to encode the intrinsic feedback from cursor movement during the execution of the response even though he is performing a secondary task. However, this feedback cannot be used during execution since the subject's attention is directed to the secondary task. Nevertheless, since this feedback may have been encoded, i t may be used during the ITI to develop the perceptual trace, as suggested by Adams (1971) and Schmidt and White (1972). Therefore, i t was considered necessary to utilize the subject's attention both during response execution and during the ITI. 60 Analysis of the Data Two dependent variables, constant error (CE) and variable error (VE), were computed from the timing measures on each subject to measure the experimental effects. In general, CE is defined as the mean algebraic error for each subject, while VE is the variability (SD) of each subject's algebraic error scores about his CE. These measures have been shown to be statistically independent (Schutz and Roy, 1973), and they have been shown to measure different behavioural phenomenon. Pepper and Herman (1970) have suggested that CE is a measure of the biasing effect of experimental treatments on the memory trace of a learned response. VE, on the other hand, is reputed to be a measure of the strength of the trace (Laabs, 1973). In this investigation, in terms of the main hypothesis, CE was used to measure the biasing effect of changing feedback following KR withdrawal on the mechanism developed during acquisition to control the response while VE was used to measure the effect of changing feedback on the strength of this mechanism to control performance. Of primary concern in this investigation were the effects following KR withdrawal. However, consideration also had to be given to the performance of all groups at the end of learning. Therefore, the last ten trials in learning for all groups were included in the analysis. Thus the 10 trials in phase one and 50 trials in phase two made a total of 60 trial measures for each subject. Constant Error. The mean algebraic error or constant error (CE) for six blocks of ten trials were calculated for each subject. These CE 61 block means were analyzed in a 2x2x3x6 (level of learning x response type x type of feedback x blocks) analysis of variance with repeated measures on the last factor. Variable Error. The variability (SD) of each subject's algebraic error scores about his mean CE within each of the six blocks was determined for each subject. The measure is referred to as variable error (VE) and was analyzed in a 2x2x3x6 (level of learning x type of response x type of feedback x blocks) analysis of variance with repeated measures on the last factor. Analysis of Intra Track Times. The intra-track times were analyzed in a 2x3x3x50 (level of learning x level of feedback x times x trials) analysis of variance with repeated measures on the last two factors. In addition, the variability (SD) of the three intra-track times was determined for each of the fifty trials for each subject. These variability measures were analyzed in a 2x3x50 (level of learning x type of feedback x trials) analysis of variance with repeated measures on the last factor. CHAPTER IV RESULTS AND DISCUSSION Results Analysis of Constant Error and Variable Error The dependent variables of CE and VE were analyzed using analysis of variance in a 2x2x3x6 (level of learning x type of response x level of feed-back x block) repeated measures factorial design (Winer, 1971: 559). For CE the mean of six blocks of ten trials formed the repeated measurement with the first block being the last ten trials in KR. For VE the varia-bility (SD) within the ten trials for each of the six blocks was the repeated block factor. A test for homogeneity of error variance and normality were carried out on the data. The two pooled within cell terms from the analysis of variance, subjects within groups (SwG) and subjects with groups by blocks (SwGxB), were partitioned into their component sum of squares (SS) for each of the twelve groups. That i s , the term SS for SwG was partitioned as follows: SS for SwGi SS for SwG12 62 63 Similarly, the term SS for SwGxB was partitioned as: SS for SwGixB SS for SwG izxB This partitioning was done for each dependent variable. These partitioned components were then tested for homogeneity using Bartlett's test (Edwards, 1960: 126). In addition, the twelve partitioned SS components were each correlated with the twelve cell means using the Spearman rank-order method. TABLE I Homogeneity and Normality of Error Variance in Data for CE and VE Measures Dependent Within Cell Data Set Measure Variance Untransformed Log Transformation X 2 P F max X 2 P F max SwG 38.55** .58* 329.09** 25.19** .39 67.62** CE SwGxB 140.76** .56* 32.09** 145.78** .48 37.67** SwG 55.49** .87* 226.65** 44.06** .69* 139.67** VE SwGxB 148.76** .89* 35.10** 103.13** .72* 21.00** *p<.05 **p<.01 64 In CE there was heterogeneity of the terms SwG (x2=38.55, p<.01) and SwGxB (x2=140.76, p<.01). In VE also there was heterogeneity of the terms SwG (x2=55.49, p<.01) and SwGxB (x2=148.76, p<.01). Also, in both CE and VE there was a significant relationship between the partitioned error terms and the cell means suggesting a slight trend toward non-normality. Since a trend toward non-normality was evident, i t was possible that Bartlett's test may have been biased in that Edwards (1960: 128) had indicated that this test is sensitive to violation of normality. Therefore, in order to corroborate the evidence provided by Bartlett's test the homo-geneity of variance was tested with the more conservative F max test. Again, however, as is indicated in Table I, this test suggested deviations from homogeneity. In order to determine i f this heterogeneity was due to a sampling bias the within cell variability for the first 15 trials of the experiment for each group were tested for homogeneity. These variances were found to be homogeneous using the F max test. Thus, the heterogeneity of variance appeared to be caused by the treatment manipulation. The heterogeneity of variance and trend toward non-normality suggested that a transformation of the data should be made. In order to decide upon what type of transformation was to be used, the range of values in CE within each of the twelve groups was determined for the untransformed data, for a square root transformation of the data and for a logarithmic transformation. The ratio of the largest range to the smallest range within the twelve groups for all three representations of the data was computed. According to Kirk (1968: 66) the transformation which is of most value is the one which provides the smallest ratio. The ratio for the untransformed data was 5.73, for the square root i t was 22.16, and for the 65 log data i t was 4.01. These results indicated that a log transformation was the more useful. Following a log transformation, the data were again subjected to an analysis of variance. The two pooled error terms were tested for homogeneity and the data were tested for normality. The results (Table I) again indicated that there was a large heterogeneity of variance as the x 2 values all exceeded the .01 level of significance. The Spearman correlation coefficient revealed that there was a smaller degree of non-normality in the data. However, the p values for VE were significant at the .05 level. After careful examination of the two data sets i t was decided that the untransformed data would be used. First, since the number of subjects per cell was equal, the analysis of variance would be fairly robust to deviations from homogeneity of error variance and normality (Edwards, 1960: 132). The reduction of heterogeneity and non-normality afforded by the log transformation was not great enough to off-set the increased difficulty in the behavioural interpretation of the data based on a logarithmic scale. Secondly, since, statistically, both analyses revealed similar trends, i t was decided that there are statistical methods, short of transformation, such as a decreased significance level, which would prevent a large increase in Type I errors in making statistical decisions. Finally, these data revealed large effects in the predicted directions which may tend to be relatively unaffected by the problems of heterogeneity and non-normality. In analyzing the results several steps were taken to reduce the probability of Type I errors. First, all main effects and interaction effects on the dependent variables of CE and VE were tested at the .01 66 level of significance. Secondly, since F max tests on several of the variances for the repeated measure, blocks, revealed heterogeneity there was some indicationof assymetry of the covariance matrices. Thus, all F ratios for within subject effects (eg. blocks, blocks by groups) were evaluated using restricted degrees of freedom (df) as suggested by Geisser and Greenhouse (in Winer, 1971: 523). This involved dividing the df for the effects of interest by (b-1), the df associated with the blocks effect. Finally, all simple effects (Winer, 1971: 347-348) and orthogonal contrasts on simple effects were tested against the specific within cell error variance associated with them, rather than with the overall pooled within cell variance (Winer, 1971: 385). This procedure effectively reduced the df over which the particular F ratio was distributed. Due to these severe restrictions all these simple effects and orthogonal compari-sons were tested at the .05 level of significance. Constant Error. The analysis of variance (Table I, Appendix B) revealed significant main effects of learning F(l ,48)=23.82, p<.01, feed-back, F(2,48)=8.03, p<.01, and blocks, F(l ,48)=9.99, p<.01. These effects are shown in Table II, These main effects must be viewed with caution as each was involved in an interaction. Nevertheless, i t was found that the LL condition exhibited significantly greater positive CE than the HL condition. In examining the effect of feedback there appeared to be a trend toward increasing positive CE from the HF to the NF condition. Post hoc orthogonal comparisons revealed that the LF condition was not significantly different from the HF condition, while the NF condition was significantly different from the combined mean of the HF and LF conditions with F(l,48)= 20.87, p<.01. A trend analysis on the main effect of blocks revealed 67 that a significant amount of variation was accounted for by a linear trend, F(l ,48)=41.96, p<.01. Although a quadratic trend was observed in this effect i t just failed significance at the .01 level. TABLE II The Main Effects in Constant Error Main Effect (seconds) Learning* Response Feedback* Blocks* HL LL B SP HF LF NF 1 2 3 4 5 6 .0053 .1318 .0638 .0733 .0232 .0413 .1413 .0052 .0494 .0785 .0851 .0941 .0992 *Effect significant, p<.01 The second order interaction of learning by type of response by type of feedback.(LRF) was significant with F(2,48)=5.48, p<.01. This inter-action is illustrated in Figure 1 (p.68) and Table I, Appendix C. In the LL condition there was a tendency for CE to show a positive increase in the LF and NF conditions relative to the HF control condition in the SP response, while in the B response only the NF condition exhibited increased CE relative to HF. In the HL condition, however, there appeared to be virtually no effect of feedback in the B response while there were large effects of manipulating feedback in the SP response. In order to more closely examine this interaction in terms of the effect of feedback on each type of response the simple main effects of F for each learning-response (L-R) combination were examined. 68 cr. o cn cn UJ F I G U R E 1 F E E D B A C K A N D L E A R N I N G E F F E C T S ON T H E S P A N D B R E S P O N S E S IN C E .300 .270 .240 .210 .180 S 150 in .120 .090 z : < £ .060 o ° .030 0 - .030 -.060 H F L F T Y P E OF F E E D B A C K © B - L L A S P - L L A S P - H L ° B - H L N F 69 Analysis of the simple main effects of feedback (Table II, Appendix B) at each L-R combination revealed that the feedback effect was not significant for the SP response in LL [F(2,12)=2.32, p>.05] or for the B response in HL (F<1). There was a significant main effect of feedback in the B response in LL [F(2,12)=5.65, p<.05] and in the SP response in HL [F(2,l2)=13.46, p<.01]. Orthogonal post hoc comparisons on each of these significant simple main effects revealed that in the SP response in HL, the NF condition was significantly different from the mean of the HF and LF conditions with F(l,12)=14.09, p<.01. The LF condition, however, was not significantly different from the HF condition, F(l,8)=3.61, p>.05. In the B-LL condition the NF condition was significantly different from HF, . F(l,8)=6.62, p<.01. This analysis revealed that the effect of feedback was different for each L-R combination. In B response in LL there was a significant increase in positive CE in the NF condition relative to the HF condition. For the SP response in LL both the NF and LF conditions exhibited large increases in positive CE. However, due to the large within cell variability in this condition these feedback effects failed significance. In HL the feedback had virtually no effect for the B response, while, for the SP response, the LF condition exhibited an increase in negative CE and the NF condition exhibited a significant increase in positive CE. Thus, this analysis suggested that the LRF interaction was due to the differences in the manner in which different types of feedback effect performance at each L-R combination. Another reason for this interaction could be due to the differences in the effect of learning at each response-feedback (R-F) combination. 69 Assessing these differences in learning effects is necessary in order to test the subsidiary hypothesis of this study concerning the effect of level of learning. Therefore, a second analysis on this LRF interaction to examine the effect of learning at each R-F combination was completed. Orthogonal post hoc comparisons at each of the three feedback conditions were made in which the LL and HL conditions were compared for each response type. The effect of learning was significant for the SP response only in the LF condition with F(l,8)=7.94, p<.05. In the B response this effect was significant only in the NF condition F(l,8)=12.68, p<.01. In addition to the LRF interaction the analysis of variance (Table I, Appendix B) revealed that there was a significant interaction between the level of learning and blocks (LB), F(l,48)=11.67, p<.01. This LB inter-action is shown in Figure 2 (p.70) and Table II, Appendix C. Analysis of this interaction (Table I, Appendix B) revealed that the interaction was composed primarily of differences in linear trend, F(l,48)=50.26, p<.01. Under the restrictions established due to heterogeneity, the differences in quadratic trend were not significant at the .01 level. Analysis of the simple main effect of blocks at each level of learning (Table III, Appendix B) revealed that the simple effect of blocks was only significant in the LL condition. A trend analysis revealed that most of the variation in blocks in this LL condition followed a linear trend, F(l,120)= 57.68, p<.01. Some of the variation, however, was accounted for by a quadratic trend [F(l ,120)=8.42, p<.01] which can be observed in the decreasing positive acceleration at about block four. A post hoc comparison revealed that divergence of the LL and HL curves became significant at block two, F(l,208)=34.79, p<.01. 70 FIGURE 2 THE BLOCKS EFFECT AT EACH LEVEL OF LEARNING .240-.220 • TRIAL BLOCKS 71 Although the main LB interaction was highly significant, i t is somewhat of a misrepresentation of the data. Closer examination revealed that the simple LB interaction affect differed, somewhat, at each response-feedback (R-F) combination leading to the suggestion of a third order LRFB interaction. Nevertheless, there was, in general, a tendency for the LL condition to exhibit increased positive CE over blocks relative to the HL condition. Variable Error. . Analysis of variance of VE (Table IV, Appendix B) showed a significant effect of level of learning with F(l,48)=11.47, p<.01. This effect can be seen in Table III. Subjects in the LL condition exhibited significantly greater variability than those in the HL condition. There was a trend for subjects in the LF and NF conditions to be more variable than those in the HF condition. However, these differences were not significant, F(2,48)=3.49, p>.01. . TABLE III The Main Effects in Variable Error Main Effect (seconds) Learning* Response Feedback Blocks HL LL B SP HF LF NF 1 2 3 4 5 6 .0704 .1012 .0806 .0909 .0719 .0843 .1012 .0776 .0918 .0898 .08*23 .0886 .0846 *Effect significant, p<.01 72 Unlike the results in CE the LRF interaction was not significant although i t did account for some of the variation with F(2,48)=l.61, p>.01. However, i t was deemed necessary to analyze the feedback effects and learning effects in this interaction, as was done in CE, for two reasons. First, an adequate test of the stated hypotheses requires that these effects be examined. Secondly, VE measures a different behavioural phenomenon than CE, that i s , VE measures the strength of the particular control mechanism developed in acquisition to exert control following the withdrawal of KR, whereas CE measures the amount of bias on the control exerted by this mechanism created by manipulating feedback in KR withdrawal. The LRF interaction in terms of VE is shown in Figure 3 (p. 73) and Table III, Appendix C. As in the analysis of CE the f i r s t analysis was carried out on the simple main effect of feedback at each L-R combina-tion (Table V, Appendix B). In the B response in LL there was a mainte-nance of VE from the HF to the LF condition and a large increase in VE in the NF condition. However, analysis of the simple main effect of feedback revealed that these changes in VE over the feedback.conditions was not significant, F(2,12)=2.18, p>.05. In the B response in HL there was an increase in VE from the HF to the LF condition followed by a decrease in the NF condition. Analysis of this simple main effect, however, revealed that these VE changes were not significant, F(2,12)=3.42, p>.05. In the SP response under both the LL and HL conditions there was a consistent increase in VE from the HF to the NF conditions. Analysis of these simple main effects showed that only in HL was this increase in VE significant, F(2,12)=24.99, p<.01. Orthogonal post hoc comparisons of this feedback effect in the SP response in HL revealed that the LF condition was signifi-cantly more variable than the HF condition, F(l ,8)=20.46, p<.01. The NF 73 .140 FIGURE 3 FEEDBACK AND LEARNING EFFECTS ON THE SP AND B RESPONSES IN VE .120 .100 d i/) g .080 DC CC Ld UJ .060 \ m < g.040 > o B-HL .020 0 HF LF TYPE OF FEEDBACK NF 74 condition, however, was not significantly different from the mean of the HF and LF conditions combined (F<1). This analysis of feedback at each L-R combination showed that the effect of feedback was significant only for the SP response in HL. Although differences in feedback conditions were observed under the other L-R combinations these were not statistically significant. The effect of level of learning in each response at each feedback condition was examined using orthogonal comparisons in which the HL and LL conditions were compared for each response under each feedback condition.' For the SP response (see Figure 3) in the HF condition there was a large difference between the HL and LL condition; however, this was not significant, F(l,8)=3.10, p>.05. From the LF condition to the NF condition the difference between the HL and LL condition showed a decrease. However, in neither the LF [F(l,8)=4.91, p>.05] nor the NF condition (F<1) was this difference significant. For the B response in HF the difference between the HL and LL conditions was significant, F(l,8)=19.87, p<.01. In the LF condition the HL-LL difference was not significant, while in the NF condition i t was significant, F(l,8)=7.06, p<.05. These results of the effect of learning revealed that, for the SP response, the learning effect decreased over the feedback conditions. However, under none of these feedback conditions was the learning effect significant. For the B response, however, a different trend was observed 75 in that the learning effect was significant under both the HF and NF conditions but not under the LF condition. Analysis of the Intra-Track Times Intra-Track Times. The intra-track times were examined in a 2x3x50x3 (level of learning x type of feedback x trials x times) complete factorial design with repeated measures on the last two factors. In this analysis a trial referred to a response and is the total of the three intra-track times that composed i t . The analysis of variance (Table VI, Appendix B) revealed a significant main effect of learning [F(l ,24)=9.12, p<.01]. Table IV reveals that the LL condition resulted in significantly longer intra-track times than the HL condition. TABLE IV The Main Effects in the Analysis of Intra-track Times Main Effects (seconds) Learning* Feedback Times HL LL HF LF NF 1 2 3 .1405 .1809 .1497 .1571 .1754 .1657 .1550 .1615 *p<.0l 76 There was a significant interaction between type of feedback and intra-track times with F(4,48)=2.70, p<.05. This interaction is depicted in Figure 4 (p.77) and Table IV, Appendix C. It can be observed that the LF and HF conditions show a trend toward reduction in time following the fir s t interval. In the NF condition, however, there is a consistent increase in the times. Post hoc comparisons of differences in time one and time three in all feedback conditions revealed that there was a significant decrease from time one to time three in the LF condition, F(l ,48)=7.46, p<.01, and a significant increase from time one to time three in the NF condition, F(l,48)=10.26, p<.01. There was no significant change in the HF condition, F(l ,48)=1.95, p>.05. Thus the reason for the interaction appears to be the difference in trend in LF and NF conditions with the LF condition showing a significant decrease and the NF condition showing a large increase. These results indicate that the subject was unable to maintain a constant velocity during the middle of the track when the feed-back was altered. Subjects in the LF condition showed a marked increase in velocity while those in the NF condition demonstrated a large decrease in velocity. Variability of Intra-Track Times. The variability (SD) of the intra-track times within each trial were examined in a 2x3x50 (level of learning x type of feedback x trials) factorial design with repeated measures on the last factor. The analysis of variance (Table VII, Appendix B) revealed a significant effect of level of learning with F(l,24)=5.38, p<.05. This effect is shown in Table V. This indicated that the subjects in the LL condition exhibited significantly greater variability in the intra-track times than those in the HL condition. No other effects achieved significance. 77 FIGURE 4 .220 i THE EFFECT OF TYPE OF FEEDBACK ON THE INTRATRACK TIMES IN AN SP RESPONSE .200 1 o 3 .180 LU a .160 LU CO CL < - 1 .140 -I A NF 2 3 INTRATRACK TIMES 78 TABLE V The Main Effects on the Analysis of Variability (SD) of Intra-track Times Main Effect (seconds) Learning* Feedback HL LL HF LF NF .0233 .0477 .0342 .0287 .0435 *p<.05 Performance on Secondary Task Performance on the secondary task in terms of per cent correct recall is indicated in Table VI. The B response showed slightly better recall than the SP response. This effect was consistent for both levels of learning. Overall, 84 per cent of the light sequences were correctly recalled. TABLE VI Per Cent Correct Recall on Secondary Task Level of Learning Type of Response Ballistic (%) Self-Paced {%) Low High 85.6 85.2 83.6 83.2 79 Summary of Main Findings Feedback- E f f e c t s . The e f f e c t of changing feedback in KR withdrawal from that which p reva i l ed during a c q u i s i t i o n dependend on the type of response involved and the amount of p r a c t i c e in a c q u i s i t i o n . When only 15 t r i a l s of p r a c t i c e were g i v e n , there tended to be increased p o s i t i v e CE and increased VE i n the LF and NF condi t ions r e l a t i v e to the HF cond i t ion in the SP response. In the B response increased CE and VE r e l a t i v e to the HF cond i t ion occurred only in the NF c o n d i t i o n . When the subjects had 150 t r i a l s of p r a c t i c e , a d i f f e r e n t trend in the data o c c u r r e d . Performance using the B response showed very l i t t l e e f f e c t of changing feedback in terms of CE. Although some increase in VE was observed in the LF cond i t ion in the B response, t h i s increase was not s i g n i f i c a n t . Performance of the SP response, however, demonstrated marked e f f e c t s of changing feedback. In terms of CE the LF cond i t ion showed a l a r g e , but non s i g n i f i c a n t , trend toward increased negative CE r e l a t i v e to the HF. c o n d i t i o n , whi le the NF cond i t ion demonstrated a s i g n i f i c a n t trend toward increased p o s i t i v e CE. In terms of i n t r a - t r a c k t imes , there was a s i g n i f i c a n t tendency f o r subjects in the NF cond i t ion to slow down over the middle of the t r a c k , whi le subjects i n the LF c o n d i t i o n showed a s i g n i f i c a n t trend toward a reduct ion in i n t r a - t r a c k t imes. The HF subjects exh ib i ted some reduct ion in times over the middle of the t r a c k , but t h i s trend was not s i g n i f i c a n t . Learning E f f e c t s . There appeared to be a general trend in CE f o r the LL cond i t ion to show a marked tendency r e l a t i v e to the HL cond i t ion toward increased p o s i t i v e CE over t r i a l s . C loser examination of the d a t a , however, revealed that the magnitude of t h i s e f f e c t was d i f f e r e n t f o r each 80 R-F combinat ion. In the SP response the LL cond i t ion was s i g n i f i c a n t l y d i f f e r e n t from the HL cond i t ion only under the LF c o n d i t i o n . In the B response t h i s learn ing e f f e c t was observed only in the NF c o n d i t i o n . With VE as the dependent measure the main e f f e c t of l eve l of l ea rn ing revealed that the subjects in the LL cond i t ion were s i g n i f i c a n t l y more v a r i a b l e than those in the HL c o n d i t i o n . C loser examination of the learn ing e f f e c t , however, showed t h a t , as in CE, th is e f f e c t was somewhat dependent on the R-F combinat ion. For the SP response, the learn ing e f f e c t was not s i g n i f i c a n t at any of the feedback c o n d i t i o n s . For the B response the l ea rn ing e f f e c t was s i g n i f i c a n t only at the HF and NF c o n d i t i o n s . D iscuss ion Impl icat ions f o r D i f f e r e n t Response Control Mechanisms The response by feedback i n t e r a c t i o n in the HL cond i t ion tends to support the f i r s t hypothesis that the e f f e c t o f changing feedback in KR withdrawal from that which p reva i l ed in a c q u i s i t i o n has a large e f f e c t on a SP response but l i t t l e e f f e c t on a B response. In developing t h i s hypo-t h e s i s the assumption was made that i f feedback was changed and CE or VE maintained t h e i r phase one values a preprogrammed contro l mechanism would be i m p l i c a t e d . A l t e r n a t e l y , i f a s u b s t a n t i a l change r e s u l t e d in the dependent v a r i a b l e s cont ingent on a l t e r i n g the feedback, a c l o s e d - l o o p mechanism would be suggested. Since changing feedback exh ib i ted r e l a t i v e l y l i t t l e e f f e c t on the performance of a B response during KR wi thdrawal , i t appears that a preprogramming mechanism developed during a c q u i s i t i o n and can be used to exp la in the contro l of performance fo l low ing the withdrawal of KR. In the SP response, however, a c l o s e d - l o o p mechanism appears to 81 have developed during acquisition since changing the feedback or preventing the use of i t tended to strongly bias the subject's response either negatively (LF condition) or positively (NF condition) as well as precipi-tating large increases in VE in the subjects' responses. Ballistic Response. A preprogrammed control mechanism seems to be strongly implicated in the performance of this B response. The availability of intrinsic feedback during the execution of the response in acquisition was minimal since an attempt was made to assure that an MT of approximately .150 seconds was involved. Previous research by Schmidt and Russell (1972) and Schmidt and White (1972) suggested that i f the MT was about .150 seconds feedback could not be used to control the response. As a result i t seems that the subject developed some type of central control mechanism which was independent of feedback. Once the ballistic response was well learned (150 trials) performance was unaffected by either changing the feedback or preventing its use. These findings conflict with those of Schmidt and White (1972) who indicated that a closed-loop mechanism explained the learning and perfor-mance of a ballistic response. They suggested that the ITI was the period where the subject used feedback to develop the perceptual trace. Since, according to closed-loop theory (Adams, 1971), feedback should be the same in KR withdrawal as that used in acquisition, then Schmidt and White (1972) would predict a change in performance in KR withdrawal i f feedback were changed. These results clearly show that no significant change in either CE or VE occurred as a result of changing the feedback. Further, even when the subject could not use the feedback, performance remained unchanged. 82 Surely such a manipulation would cause a performance change i f a closed-loop mechanism, as suggested.by Schmidt and White (1972), were operating. Although the manipulations of feedback had very l i t t l e effect on the performance of the well learned B response during KR withdrawal, i t is important to consider how feedback was changed. Marteniuk and Roy (1973) have suggested that i f feedback is changed, by changing the characteristics of the task (eg. proprioceptive cues), as was done in the LF condition, a decrement in a preprogrammed response would be predicted. This decrement should have been caused not so much by the feedback having been changed, but rather by the fact that the characteristics of the task, which were used to develop the motor program, had been changed. Thus, the motor program developed during acquisition would not be adequate to maintain performance during KR withdrawal since the task demands, and not just the feedback, had been changed. These data, in general, do not support this suggestion since no significant change in performance of the B response occurred in the LF condition. Nevertheless, there was a marked tendency for subjects in the LF condition to be more variable than in either of the other two feedback conditions. This observation suggests that VE may be more sensitive than CE to the change predicted by Marteniuk and Roy. That is, changing the demands of the task may not tend to bias the response, as measured by CE, rather, i t may tend to weaken the control exerted by the motor program. This weaker control could be exhibited in VE in the same way as Laabs (1973) uses VE to indicate the weakening of a memory trace. Self-Paced Response. Two lines of evidence strongly suggest that a closed-loop mechanism was developed in the SP response during acquisition 83 and was subsequently used to control performance following KR withdrawal. First, there were large effects of manipulating feedback in this response. In the HL condition where enough practice was given to fully develop the response mechanism performance in the NF condition showed a trend toward increased positive CE and increased VE, while performance in the LF condi-tion showed an increase in negative CE and increased VE. No such large effects of feedback manipulation should have occurred i f a preprogramming mechanism had developed (cf. B response). Secondly, the intra-track data indicated that there were large differences in the intra-track times. Subjects in the NF condition showed a trend toward decreased speed as evidenced by an increase in the intra-track times. Those subjects in-the HF and LF conditions showed a marked tendency to increase their velocity in the middle of the track as indicated by a decrease in the intra-track times. Again, such large changes in the intra-track times would not be predicted i f a preprogrammed mechanism was involved because, once the program was fired, the velocity over the middle of the track would be quite constant since the subject would not be altering his response based on feedback. In general, these findings concerning the effects of changing feedback in KR withdrawal on an SP response concur with those of Adams, Goetz and Marshall (1972) as these authors, using absolute error (AE) as the dependent measure, found a significant increase in AE when feedback was changed. It is not possible to compare the CE and VE findings with Adams' AE data much further than this as i t is uncertain whether Adams' AE data reflect changes in CE, VE or both since Schutz and Roy (1973) have shown AE to be a function of CE and VE. 84 Finally, in terms of the main purpose of this study these results clearly indicate that the type of theory used to explain the learning and control of performance depends on the type of response. When feedback was available to guide the response (SP response), then the subject used the feedback and developed a closed-loop or feedback dependent control mechanism in acquisition. When feedback was not available to guide the response (B response), i t appeared that the subject developed a more central performance control mechanism in acquisition which was not dependent on feedback. Feedback Manipulations Following High and Low Learning High Learning. Hypothesis 1 which stated that changing feedback or interrupting feedback results in a change in CE and VE in an SP response but not in B response was supported by these data. Feedback manipulation had l i t t l e effect on the B response suggesting that a preprogrammed mechanism independent of feedback had developed. In the SP response, however, large changes in CE and VE occurred suggesting that a closed-loop mechanism had developed in acquisition. Low Learning. Hypothesis 2 stating that all feedback conditions would result in equally poor performance in CE and VE in both an SP and B response was not generally supported by these data. For the B response there was a tendency for the HF and LF conditions to be the same in terms of both CE (Figure 1) and VE (Figure 3), while the NF condition showed a marked increase in both CE and VE. The observation that changing feedback had no marked effect on performance relative to the control (HF) condition suggests that some type of primitive motor program may have developed in the B response after only 15 trials. This programmed mechanism may be 85 unaffected by changes in feedback. A l t e r n a t e l y , the B response may have been such that even the small amount of feedback provided in the LF c o n d i t i o n was enough to maintain performance. Indeed, t h i s i s s t rong ly suggested by the f a c t that when feedback was unava i lab le to the subject (NF cond i t ion ) a la rge decrement in performance r e s u l t e d . For the SP response the increase in CE and VE pred ic ted by hypothesis 2 occurred in the LF and NF c o n d i t i o n s . However, th is increase r e l a t i v e to HF was not s t a t i s t i c a l l y s i g n i f i c a n t . In terms of CE (Figure 1) a la rge increase in the p o s i t i v e d i r e c t i o n r e l a t i v e to the HF cond i t ion r e s u l t e d in both the LF and NF c o n d i t i o n . For VE (Figure 3) a large increase in VE r e s u l t e d in both the LF and NF c o n d i t i o n s . Even though these trends were not s t a t i s t i c a l l y s i g n i f i c a n t they were comparable to the s i g n i f i c a n t changes which resu l ted in the LF and NF condi t ions f o r the SP response under HL in both CE and VE. It i s i n t e r e s t i n g to compare these e f f e c t s of feedback in LL between response types . S ince a decrement in performance r e l a t i v e to the HF c o n d i t i o n occurred in the NF c o n d i t i o n in both the SP and B response, feedback appeared to be necessary f o r both response types . However, s i n c e the e f f e c t of changing feedback (LF) was evident only in the SP response, the amount of feedback requi red to maintain performance appeared to be d i f f e r e n t . Whereas a small amount of feedback was necessary f o r the B response to maintain performance, a l a rger i n t e n s i t y of feedback appeared to be necessary to maintain performance i n the SP response. This comparison seemed to i n d i c a t e t h a t , even a f t e r 15 t r i a l s of p r a c t i c e in a c q u i s i t i o n , there appeared to be some d i f f e r e n t i a t i o n between the mechanisms used to cont ro l performance in each type of response. 86 Effect of Learning on Performance During KR Withdrawal Self-Paced Response. Hypothesis 3 stating that the effect of learning in the SP response is evident only in the HF condition was not supported by the CE data. The difference between the HL and LL conditions was only significant under the LF condition. This appears to be due to the -fact that the LF condition affected the direction of error differently for this response at each level of learning. In LL there was a marked tendency toward a positive shift in CE, while in HL a shift in the negative direction occurred. It is uncertain why changing the characteristics of feedback (LF) had a different effect at each level of learning. Possibly the biasing effect of this feedback condition has a different effect when the habit strength of the perceptual trace is low (LL) than when the strength of the trace is greater (HL). This hypothesis finds some support in examining VE, a measure of the strength of a memory trace, under this LF condition (Figure 3). The variability of the subjects' responses was greater in the LL than in the HL condition suggesting that the strength of the trace was less in the LL condition. As indicated above i t appeared that when CE was the dependent measure no systematic trend as was predicted by hypothesis 3 in the effect of level of learning occurred due, probably, to the differential biasing effects of the feedback conditions at each level of learning. With VE as the depen-dent measure, however, a more orderly trend in the learning effects in the direction predicted by hypothesis 3 was seen. The mean difference between the HL and LL condition decreased from HF to the NF conditions. The mean difference in HF was .0398, in LF i t was .0248, and in NF i t was .0135. In all cases the greater amount of variability was in the LL condition. 87 Closer examination of the data for the SP response (Figure 3) revealed that this decreasing effect of learning over feedback conditions resulted from a differential rate of increase in VE under each learning condition over the three feedback conditions. In LL there was a small increase in VE from the HF to the NF condition. In HL, however, a much greater increase in VE resulted from the HF to the NF condition. These results in VE can be interpreted in terms of the strength of the perceptual trace for controlling performance. When the feedback during KR withdrawal was the same as that used during acquisition (HF), providing the subject with 150 trials (HL) as compared with 15 trials (LL) had a marked effect on the strength of the perceptual trace. Since the VE was substantially smaller in the HL than the LL, the strength of the perceptual trace appeared to be much greater in the HL condition. When feedback was changed during KR withdrawal from that which was present during acquisition (LF) or when the subject was unable to use feedback during this period (NF), the difference, in terms of VE, between providing the subject with 15 or 150 trials was much smaller than when feedback was unchanged. This seems to indicate that the strength of the perceptual trace for controlling performance was markedly reduced in HL condition under both altered feedback conditions and approached the trace strength exhibited when only 15 trials in practice were given. It is interesting to compare these results in CE and VE with the findings of Adams, Goetz and Marshall (1972) who used absolute error (AE) as the dependent measure. Adams, Goetz and Marshall (1972) found that the effect of the level of learning in acquisition on performance following the withdrawal of KR depended on whether feedback during KR withdrawal was 88 the same as that in acquisition. If the feedback was the same, there was significantly greater error in the group which only had 15 trials in acquisition compared with the high learning group (150 trials). If, however, feedback was not the same, no difference in absolute error was found between the two learning groups. They explained these results in the following manner. When feedback was the same as that in acquisition, the strengthened perceptual trace in the HL condition enabled the subject to maintain performance with less error than in the LL condition in which the trace was much weaker. When feedback was not the same, however, the strengthened perceptual trace in the HL condition did not afford the ability to maintain performance in KR with-drawal because the perceptual trace was unable to utilize the altered feedback to maintain performance. Therefore, the HL condition would be expected, as they found, to exhibit as much error as the LL condition. As was mentioned above, i t is somewhat difficult to directly compare the findings of this study with those of Adams, Goetz and Marshall (1972) because the dependent variable, AE, which they used, is a function of the two dependent variables, CE and VE, used in this study (Schutz and Roy, 1973). Nevertheless, i f the Adams, Goetz and Marshall (1972) data primarily reflected changes in VE then the data in this study would support their findings. Therefore, i t appears that the learning effect discussed by Adams, Goetz and Marshall (1972) was evident in this data only when the variability of the subjects' responses were considered. If one considers VE as a measure of the strength of a memory trace, these results appear logical. When a great deal of practice was given and feedback was unchanged, the strength of the trace was considerably greater in the HL than in the 89 LL condition. However, when feedback was changed, a significant increase in the variability of the subjects' responses resulted in the HL condition suggesting that the perceptual trace was unable to use the feedback to maintain performance. Consequently, the difference between HL and LL decreased when the feedback was changed. Ballistic Response. Hypothesis 4 stating that the effect of learning would be significant at all levels of feedback in the B response was not supported by the data in either CE or VE. In CE (Figure 1) there was a tendency for the difference between the HL and LL conditions to remain fairly constant in the HF and LF conditions and significantly increase in the NF condition. In VE (Figure 3) the HL-LL difference was significant in HF, decreased in the LF condition and increased signifi-cantly in the NF condition. In comparing the findings in CE and VE there was a trend in both for the HL-LL difference to be significant in the NF condition. This finding resulted from the fact that the preventing the subject from using feedback (NF) had a marked effect on the B response in LL but very l i t t l e effect in HL. This NF result coupled with the fact that the LF condition did not result in a significant change in performance relative to the HF condition in either HL or LL suggested that, in both levels of learning, some type of preprogrammed mechanism may have developed. In LL a primitive type of programmed mechanism may have developed while in HL a more sophisticated mechanism seemed to be present. The strength of the mechanism developed after 15 trials appeared not to be as great in exerting control over the performance of the B response as the one 90 developed after 150 trials of practice in acquisition since, in the control condition (HF), a significant difference in VE was evidenced. Further, i t appeared that this more primitive programmed mechanism, although not affected by changing the characteristics of feedback (LF), was greatly affected by preventing the subject from using feedback (NF). It appeared, therefore, that the mechanism controlling performance was s t i l l somewhat dependent on feedback. Possibly, the standard in the central loop had not developed to a sufficient degree to enable this loop to operate. These data suggest that progressing from 15 to 150 trials of practice in acqui-sition may have marked a transition from a rather primitive programmed mechanism in which the central control loop was 'not fully formed to a more sophisticated mechanism in which the central loop was fully functional. Attention Demands Keele (1968) has suggested that one of the advantages of pre-programming is that the amount of attention paid to performance can be greatly reduced. In examining the effect of the NF condition in HL i t appears that such a reduction of attention demand has occurred in the B response as a result of preprogrammed control. When the subject's attention was occupied with the secondary task in the B response, no appreciable effect either in response biasing (CE) or intra-variance (VE) occurred. This was attributed to the development of preprogrammed control in acquisition. In the SP response, however, the NF condition effected a significant positive bias in performance and significantly increased the variability of performance. This effect was attributed to the development of closed-loop control in the SP response during acquisition. 91 Finally, the fact that there was very l i t t l e difference between the SP and B responses in HF, where no secondary attention demanding task was present indicated that both types of responses were equally effective in maintaining performance following KR withdrawal. However, the amount of attention required to maintain this performance was much less in the B response since, when attention was occupied (NF), no decrement in perfor-mance resulted. Thus, these findings tend to concur with Keele's suggestion that preprogramming may be a more efficient control mechanism due to the reduced attention demand. It should be noted, however, that i t is not suggested that preprogramming requires no attention since Posner (1969) has shown that performance which is automated, such as in a preprogrammed response, does require some amount of central processing capacity. Rather, it is suggested that, since the subject using the B response was not continually required to compare intrinsic feedback to the perceptual trace as in a closed-loop mechanism, a reduction in the attention demand of the task resulted. Although i t appears here that a greater degree of attention is required in the SP response because of the necessity to compare feedback to the perceptual trace, i t is possible that i f many more trials had been given on the SP response the degree of attention required might have been reduced considerably. Adams (1971) suggested that this comparison of feedback to the perceptual trace may become almost "automatic" after thousands of trials, thus reducing the attention load. Although this is possible, i t is equally likely that as practice continues the subject may actually require less feedback for comparison with the perceptual trace than was required in the early stages of performance in KR withdrawal as he may be able to anticipate the total amount of feedback information he will 92 receive on any response from only the fi r s t sampling of that information. Once he has received the f i r s t sample of the feedback information the rest may be redundant and not require much attention to process. Motor Memory Recently, Pepper and Herman (1970) have proposed a theory of motor memory which uses CE as a measure of forgetting. They have shown, f i r s t , that the intensity of feedback of an interpolated motor activity tends to bias the stored response in the direction of the intensity of feedback from the interpolated task. For instance, i f the proprioceptive feedback of an interpolated motor response was less than the feedback associated with the learned response CE on recall of the learned response showed a marked negative trend. Secondly, they have shown that occupying the subject's attention during the retention interval with some form of non-motor task tended to cause a positive shift in CE in the recall of the learned response. The data for the SP response in HL tends to support the findings of Pepper and Herman (1970). In the LF condition where the intensity of the proprioceptive feedback was reduced, there was a marked tendency toward negative CE. In the NF condition when the subject's attention was occupied with a non-motor task, a positive bias in CE occurred. Both these results would be predicted from the work of Pepper and Herman (1970). 93 CHAPTER V SUMMARY AND CONCLUSIONS Summary The main purpose of this study was to determine the generality of the closed-loop theory and the preprogramming theory as an explanation for learning the maintenance of performance in a highly practiced self-paced (SP) and ballistic (B) response. The method used for determining which of these theories best explained the learning and maintenance of performance in each response involved comparing performance, following the withdrawal of KR, under changed or interrupted feedback conditions to a control condition in which feedback was the same as that used in acquisition. Other problems investigated in this study were (1) the effects of changing or interrupting feedback during KR withdrawal following low practice in acquisition and, (2) the differential effect of high and low practice in acquisition on performance in each response type in KR withdrawal when feedback was the same as in acquisition, when feedback was different from that in acquisition, and when feedback was interrupted. Sixty (36 males, 24 females) right handed graduate and undergraduate students of the University of British Columbia were used as subjects. Two males and three females were assigned to one of twelve groups. The experimental task involved learning to move a cursor on a track from one end of the track to the other in 1.0 seconds. Two types of responses were used: self-paced (SP) and ballistic (B). In the SP response the 94 subject was permitted to hold on to the cursor for the entire length of the track. In the B response the subject had to release the cursor after he moved i t only about one sixth of the track distance. Thus he had to push the cursor fast enough in the f i r s t one sixth of the track so that i t would get to the other end in the required 1.0 second interval. The experiment was divided into two phases. Phase one was the learning phase and involved four groups which were designated according to whether the subjects used a B or a SP response and whether they had 15 (low learning or LL) or 150 trials (high learning or HL) of practice. All groups learned under high feedback and KR from the experimenter. Phase two of the experiment involved 50 trials without external KR. Each of the four groups of phase one was divided into 3 feedback groups: high feedback (HF) in which subjects performed the timing response as in phase one, low feedback (LF) in which the intensity of proprioceptive and auditory feedback was reduced, and no feedback (NF) in which feedback was interrupted by having the subject perform a secondary task. The results indicated that, for a highly practiced B response, changing feedback or interrupting feedback had no effect on performance relative to the control condition duirng KR withdrawal, while these feed-back manipulations resulted in a significant change in CE and VE relative to the control condition in a highly practiced SP response. Following LL in acquisition changing feedback or interrupting feedback in KR withdrawal resulted in a large change in CE and VE relative to the control condition in the SP response, while, for the B response only interrupting feedback resulted in a significant change in CE and VE relative to the control. 95 F i n a l l y , fo r the SP response the d i f fe rence , in terms of CE, between providing the subject with HL or LL in acqu is i t i on was s i g n i f i c a n t only in the LF cond i t ion . In terms of VE, the HL-LL d i f ference was not s i g n i f i c a n t under any feedback cond i t ion ; however, there was a trend toward a decrease in th is d i f ference from the HF to the LF and NF condi t ions. For the B response, on the other hand, the HL-LL d i f fe rences , in terms of CE, were s i g n i f i c a n t only in the NF cond i t ion , wh i l e , in terms of VE, the HL-LL d i f ference was s i g n i f i c a n t under the HF and NF condi t ions. Conclusions The conclusions were: 1. That the preprogramming theory explained the learning and maintenance of performance in the highly pract iced B response, whi le the c losed-loop theory explained learning and performance in the h ighly prac t iced ' SP response. 2. That fo l lowing a small amount of p rac t ice in acqu is i t i on both response types were dependent on feedback. However, the amount of feedback needed to maintain performance was much less in the B response than in the SP response. 3. That performance fo l lowing a small amount of pract ice in acqu is i t i on (LL) as compared with performance fo l lowing a large amount of pract ice in acqu is i t i on (HL) in the B response indicated that there was a t rans i t i on from a pr imi t ive preprogrammed mechanism which was somewhat 96 d e p e n d e n t on f e e d b a c k t o a w e l l d e v e l o p e d p r e p r o g r a m m e d m e c h a n i s m w h i c h was n o t d e p e n d e n t on f e e d b a c k . 4. T h a t a c l o s e d - l o o p m e c h a n i s m f o r p e r f o r m a n c e c o n t r o l was s u g g e s t e d f o r t h e SP r e s p o n s e f o l l o w i n g b o t h s m a l l a n d l a r g e a m o u n t s o f p r a c t i c e i n a c q u i s i t i o n . BIBLIOGRAPHY ADAMS, J.A., Human Memory. New York: McGraw-Hill Co., 1967. ADAMS, J.A., Response feedback and learning. Psychological Bulletin, 70:486-504, 1968. ADAMS, J.A., A closed-loop theory of motor learning. Journal of Motor  Behavior, 3:111-149, 1971. ADAMS, J.A. and BRAY, N.W., A closed-loop theory of paired associates verbal learning. Psychological Review, 77:385-405, 1970. ADAMS, J.A., GOETZ, H.T. and MARSHALL, P.H., Response feedback and motor learning. Journal of Experimental Psychology, 92:391-397, 1972. ADAMS, J.A., MARSHALL, P.H. and GOETZ, E.T., Response feedback and short-term motor retention. Journal of Experimental Psychology, 92:92-97, 1972. ANGEL, R.W. and GARLAND, H., Transfer of information from manual to oculomotor control system. Journal of Experimental Psychology, 96:92-96, 1972. ANOKHIN, P.K., Cybernetics and the integrative activity of the brain. In M.E. Cole and I. Maltzman (Eds.), A Handbook of Contemporary  Soviet Psychology. New York: Basic Books, 1969, Pp. 830-856. BAHRICK, H.P. and BAHRICK, P.D., A re-examination of the interrelations among measures of retention. Quarterly Journal of Experimental  Psychology, 16:318-324, 1964. BECK, E.C. and DOTY, R.W., Conditioned flexion reflexes acquired during combined catalepsy and de-efferentation. Journal of Comparative  and Physiological Psychology. 50:211-216, 1957. BIL0DEAU, I.McD., Information feedback. In E.A. Bilodeau (Ed.), Principles  of Skill Acquisition. New York: Academic Press, 1969, Pp. 255-285. CHANG, H.T., Activation of internuncial neurons through collaterals of pyramidal fibres at the cortical level. Journal of Neurophysiology, 18:452-471, 1955. CHASE, R.A., An information flow model of the organization of motor activity I: Transduction, transmission and central processing of sensory information. Journal of Nervous and Mental Disease, 140:239-251, 1965(a). 0"7 98 CHASE, R.A., An information flow model of the organization of motor activity II: Sampling, central processing and utilization of sensory information. Journal of Nervous and Mental'Disease, 140:239-251, 1965(b). CHASE, R.A., HARVEY, S., STANDFAST, S. RAPIN, I., and SUTTON, S., Studies on sensory feedback I: Effect of delayed auditory feedback on speech and keytapping. Quarterly Journal of Experimental Psychology, 13:141-152, 1961. CHASE, R.A., RAPIN, I., GILDEN, L., SUTTON, S., and GUILFOYLE, G., Studies on sensory feedback II: Sensory feedback influences on keytapping motor tasks. Quarterly Journal of Experimental Psychology, 13:153-167, 1961. CHASE, R.A., SUTTON, S., and RAPIN, I., Sensory feedback influences on motor performance. Journal of Auditory Research, 3:212-223, 1961. CHERNIKOFF, R. and TAYLOR, F.V., Reaction time to kinesthetic resulting from sudden arm displacement. Journal of Experimental Psychology, 43:1-8, 1952. CRAIK, K.J.W., Theory of the human operator in control systems I: The operator as an engineering system. British Journal of Psychology, 38:56-61, 1947. CRAIK, K.J.W., Theory of the human operator in control systems II: Man as an element in a control system. British Journal of Psychology, 38:142-148, 1948. CRAIK, K.J.W., The operator as an engineering system. In D. Legge (Ed.), Skills. Middlesex, England: Penguin Books Ltd., 1970, Pp. 148-156. DAVIS, R., SUTHERLAND, N.S. and JUDD, B.R., Information content in recognition and recall. Journal of Experimental Psychology, 61:422-429, 1961. DEWEY, J., The reflex arc concept in psychology. Psychological Review, 3:357-370, 1896. EDWARDS, A.L., Experimental Design in Psychological Research. New York: Holt, Rinehart & Winston, 1960. ELLIS, M.J., Control dynamics and timing a motor response. Journal of  Motor Behavior, 1:119-134, 1969. EIMAS, P.D. and ZEAMAN, D., Response speed changes in an Estes' paired-associate "miniature" experiment. Journal of Verbal Learning and  Verbal Behaviour, 1:384-388, 1963. FESTINGER, L. and CANON, L.K., Information about spatial location based on knowledge about efference. Psychological Review, 72:373-384, 1965. 99 FESTINGER, L., ONO, H., CLARKE, A.B., and BAMBER, D., Efference and the conscious experience of perception. Journal of Experimental  Psychology, Monograph Supplement, 74: Whole No. 637, 1967. FITTS, P.M. and POSNER, M.I., Human Performance. Belmont, California: Wadsworth Co., 1969. FLEISHMAN, E.A. and HEMPEL, W.E., Changes in factor structure of a complex psychomotor test as a function of practice. Psychometrika, 19:239-252, 1954. GIBBS, C.B., Thecontinuous regulation of skilled response by kinesthetic feedback. British Journal of Psychology, 45:24-39, 1954. GREENWALD, A.G., Sensory feedback mechanisms in performance control: With special reference to the ideo-motor mechanism. Psychological  Review, 77:73-99, 1970. GREENWALD, A.G. and ALBERT, S.M., Observational learning: A technique for elucidating S-R mediation processes. Journal of Experimental  Psychology, 76:267-272, 1968. HELD, R., Exposure history as a factor in maintaining stability of perception and coordination. Journal of Nervous and Mental Disease, 132:26-32, 1961. HIGGINS, J.R. and ANGEL, R.W., Correction of tracking errors without sensory feedback. Journal of Experimental Psychology, 84:412-416, 1970. HOLST, E.von., Relations between the central nervous system and the peripheral organs. British Journal of Animal Behaviour, 2:89-94, 1954. HONZIK, C.H., The sensory basis of maze learning in rats. Comparative  Psychological Monographs, 13: Whole No. 64, 1936. HULL, C.L., Knowledge and purpose as habit mechanisms. Psychological Review, 37:511-525, 1930. HULL, C.L., Goal attraction and directing ideas concerned as habit phenomena. Psychological Review, 38:487-506, 1931. IRVINE, S.R. and LUDVIGH, E., Is ocular proprioceptive sense concerned in vision? A.M.A. Archives of Ophthalmology, 15:1037-1049, 1936. JAMES, W., Principles of Psychology. Volume I. New York: Holt, 1890(a). JAMES, W., Principles of Psychology. Volume II. New York: Holt, 1890(b). JONES, J.E., All or none versus incremental learning. Psychological  Review, 69:156-160, 1962. KEELE, S.W., Movement control in skilled motor performance. Psychological  Bulletin, 70:387-403, 1968. 100 KEELE, S.W. and POSNER, M.I., Processing of visual feedback in rapid movements. Journal of Experimental Psychology, 77:155~1583 1968. KIRK, R.E., Experimental Design: Procedures for the Behavioural Sciences. California: Brooks/Cole Co., 1968. KUYPERS, H.G.J.M., Central coritical projections to motor and somato-sensory cell groups. Brain, 83:161-184, 1960. LAABS, G.E., Retention characteristics of different reproduction cues in motor short-term memory. Journal of Experimental Psychology, 1973, in press. LASHLEY, K.S., The accuracy of movement in the absence of excitation from the moving organ, American Journal of Physiology, 43:169-194, 1917. LASHLEY, K.S., The problem of serial order in behaviour. In L.A. Jeffress (Ed.), Cerebral mechanisms in Behaviour. New York: Wiley, 1951, Pp. 112-136. LASHLEY, K.S. and BALL, J., Spinal conduction and kinesthetic sensitivity in the maze habit. Journal of Comparative Psychology, 9:71-106, 1929. LASHLEY, K.S. and MCCARTHY. D.A., The survival of the maze habit after cerebellar injuries. Journal of Comparative Psychology, 6:423-434, 1926. LASZLO, J.I., The performance of a simple motor task with kinaesthetic sense loss. Quarterly Journal of Experimental Psychology, 18:1-8, 1966. LASZLO, J.I., Training fast tapping with reduction of kinesthetic, tactile, visual and auditory sensations. Quarterly Journal of Experimental  Psychology, 19:344-349, 1967(a). LASZLO, J.I., Kinesthetic and exteroceptive information in the performance of motor skills. Physiology and Behaviour, 2:359-365, 1967(b). LASZLO, J.I. and MANNING, L.C, The role of motor programming, command and standard in the central control of skilled performance. Journal  of Motor Behaviour, 2:111-124, 1970. LASZLO, J.I., SHAMOON, J.S. and SANSON-FISCHER, R.W., Reacquisition and transfer of motor skills with sensory feedback reduction. Journal  of Motor Behaviour, 1:195-209, 1969. LI, CH-L., Activity of interneurons in the motor cortex. In H.H. Jasper, L.D. Proctor, R.S. Knighton, W.C. Norsbay and R.T. Costello (Eds.), Reticular Formation of the Brain. Boston: Little, Brown & Co., 1958, Pp. 459-472. LUH, C.W., The conditions of retention. Psychological Monographs, 31: No. 142, 1922. 101 MARSHALL, P.H., Recognition and recall in short-term motor memory. Journal of Experimental Psychology, 95:147-153, 1972. MARTENIUK, R.G. and ROY, E.A., Control mechanisms in motor performance. In preparation, 1973. MERTON, P.A., Human position sense and sense of effort. Society of Experimental Biology Symposium XVIII, Homeostasis and Feedback Mechanisms: Cambridge: Cambridge University Press, 1964, Pp. 387-400 (a). MERTON, P.A., Absence of conscious position sense in the human eyes. In M.B. Bender (Ed.), The Oculomotor System. New York: Harper & Row, 1964, Pp. 314-320~[b7: MILLER, G.A., GALANTER, E. and PRIBRAM, K.H., Plans and The Structure of  Behaviour. New York: Holt, Rinehart & Winston, 1960. NOBLE, M., TRUMBO, D., and FOWLER, F., Further evidence on secondary task interference in tracking. Journal of Experimental Psychology, 73:146-149, 1967. PAILLARD, J. and BROUCHON, M., Active and passive movements in the c a l i -bration of position sense. In S.J. Freedman (Ed.), The Neuropsychology of Spatially Oriented Behaviour. Illinois: The Dorsey Press, 1968, Pp. 37-56. PAVLOV, I.P., A physiologist's answer to psychologists. In Twenty Years' Experience in the Objective Study of the Higher Nervous Activity  (Behaviour) of Animals. Leningrad, 1932. PEPPER, R.L. and HERMAN, L.M., Decay and interference effects in the short-term retention of a discrete motor act. Journal of Experimental  Psychology, 83:2, Pt. 2, 1970. POSNER, M.I., Reduced attention and the performance of "automated" movements. Journal of Motor Behaviour, 1:245-258, 1969. POSNER, M.I. and BOIES, S.J., Components of attention. Psychological Review, 78:391-408, 1971. POSNER, M.I. and KEELE, S.W., Time and space measures of mental operations. Paper presented to American Psychological Association, September, 1970. PROVINS, K.A., The effect of peripheral nerve block on the appreciation and execution of finger movements. Journal of Physiology, 143:55-67, 1958. RABBITT, P., Error correction without external error signals. Nature, 212:488, 1966. RABBIT, P., Time to detect errors as a function of factors affecting choice-response time. Acta Psychologica, 27:131-142, 1967. 102 SCHMIDT. R.A., Movement time as a determiner of timing accuracy. Journal  of Experimental Psychology, 79:43-47, 1969. SCHMIDT. R.A., The index of preprogramming (IP): A statistical method for evaluating the role of feedback in simple movements. Ps.ychonomic  Science, 27:83-85, 1972. SCHMIDT, R.A. and RUSSELL, D.G., Movement velocity and movement time as determiners of the degree of preprogramming in simple movements. Journal of Experimental Psychology, 1972, in press. SCHMIDT. R.A. and WHITE, J.L., Evidence for an error detection mechanism in motor skills: A test of Adams' Closed-Loop Theory. Journal of  Motor Behaviour, 1972, in press. SCHUTZ, R.W. and ROY, E.A., Absolute error: The devil in disguise. . Journal  of Motor Behaviour, 1973, in press. SHERRINGTON, C.S., The Integrative Action of the Nervous System. New York: Scribner, 1906. SMITH, K.U., Cybernetic theory and analysis of learning. In E.A. Bilodeau (Ed.), Acquisition of Skill. New York: Academic Press, 1966, Pp. 425-482. S0K0L0V, E.N., The modelling properties of the nervous system. In M.E. Cole and I. Maltzman (Eds.), A Handbook of Contemporary Soviet Psychology. New York: Basic Books, 1969, Pp. 671-704. SOLOMON, R.L. and TURNER, L.H., Discriminative classical conditioning in dogs paralyzed by curare can later control discriminative avoidance responses tithe normal state. Psychological Review, 69:202-219, 1962. SPERRY, R.W., Neural basis of the spontaneous optokinetic response produced by visual neural inversion. Journal of Comparative and Physiological  Psychology, 43:482-489, 1950. TAUB, B. and BERMAN, A.J., Avoidance conditioning in the absence of relevant proprioceptive and exteroceptive feedback. Journal of Comparative  and Physiological Psychology, 56:1012-1016, 1963. TAUB, B. and BERMAN, A.J., Movement and learning in the absence of sensory feedback. In S.J. Freedman (Ed.), The Neuropsychology of Spatially  Oriented Behaviour. Illinois: The Dorsey Press, 1968, Pp. 173-192. TAUB, B., BACON, R., and BERMAN, A.J., The acquisition of trace-conditioned avoidance response after deafferentation of the responding limb. Journal of Comparative and Physiological Psychology, 58:275-279, 1965. TOLMAN, E.C, Prediction of vicarious trial and error by means of the schmatic sowbug. Psychological Review, 46:318-336, 1939. 103 TROLAND, L.T., The Fundamentals of Human Motivation. New York: Van Nostrand, 1928. WASHBURN, M.F., Movement and Mental Imagery. New York: Houghton Mifflin, 1916. WATSON, J.B., Behaviorism. Chicago: University of Chicago Press, 1930. WEINER, N., Cybernetics. New York: Wiley, 1948. WELFORD, A.T., Fundamentals of Ski l l . London, England: Metheun, 1968. WILLIAMS, LD. and ROY, E.A., Closed-loop control of a ballistic response. Journal of Motor Behavior, 4:121-126, 1972. WILSON, D.M., The central nervous control of flight in a locust. Journal  of Experimental Biology, 38:471-490, 1961. WILSON, D.M., The origin of the flight-motor command in grasshoppers. In R.F. Reiss (Ed.), Neural Theory and Model ling. Stanford: Stanford University Press, 1964, Pp. 331-345. WILSON, D.M., Inherent assymetry and reflex modulation of the locust flight pattern. Journal of Experimental Biology, 48:631-641, 1968. WILSON, D.M. and GETTRUP, E., A stretch reflex controlling wingbeat frequency in grasshoppers. Journal of Experimental Biology, 49:171-185, 1963. WILSON, D.M. and WYMAN, R.J., Motor output patterns during random and rythmic stimulation of locust thoracic ganglia. Biophysical Journal, 5:121-143, 1965. WINER, B.J., Statistical Principles in Experimental Design. New York: McGraw-Hill Co., 1971. WISDOM, J.O., The hypothesis of cybernetics. British Journal for the  Philosophy of Science, 2:1-24, 1951. YATES, A.J., Delayed auditory feedback. Psychological Bulletin, 60:213-232, 1963. 104 APPENDIX A Apparatus 105 FIGURE 1 Experimental Apparatus 106 FIGURE 2 Distances Between Contact Points of the Microswitches in the Track M 8 M 7 M 6 M 5 M - M3 M 2 Mj —I — 1 i I I I l ,—t— |< 6-c" 1 ' S " _ > | | * — - S"—> | | * — «• - 5" > ( < 6.0" >| ITT 3 ITT 2 ITTi 27-5 Total Timed Track Distance 3"e o Total Track Distance 107 APPENDIX B S t a t i s t i c a l A n a l y s e s 108 TABLE I Analysis of Variance of Constant Error Source of Variance df MS Between Subjects Learning (L) 1 Response (R) 1 Feedback (F) 2 LxR 1 LxF 2 RxF 2 LxRxF 2 SwG 48 1.440709 .008161 .485514 .008703 .153269 .292325 .331509 .060491 23.8* <1 8.03* <1 2.53 <1 5.48* Within Subjects Blocks (B) B(linear) B(quad) LxB LxB(linear) LxB(quad) .076241 .088998 .320241 ,054313 .383612 .048270 9.99* 41.96* 7.12 11.66* 50.26* 6.32 RxB FxB LxRxB LxFxB RxFxB LxRxFxB SwGxB 5 10 5 10 10 10 240 .004479 .031718 .046088 .219099 .088162 .268715 .007633 <1 4.16 <1 2.87 1.16 3.21 *p<.01 109 TABLE II Analysis of Variance for the Simple Main Effect of Feedback At Each Learning-Response Combination in Constant Error Source of Variance df MS F F at HL-B 2 .003971 <1 F at HL-SP 2 .228252 13.46* F at LL-B 2 .553489 5.65* F at LL-SP 2 .213833 2.32 SwHL-B 12 .034995 SwHL-SP 12 .016961 SwLL-B 12 .097993 SwLL-SP 12 .092015 *p<.05 n o TABLE III Analysis of Variance for Simple Main Effect of Blocks at Each Level of Learning in Constant Error Source of Variance df MS Blocks at HL 5 Blocks at LL 5 Blocks at LL(linear) Blocks at LL(quad) SwHL 120 SwLL 120 .000548 .164701 .003089 .012179 .702427 ,102555 <1 13.52* 57.68* 8.42* *p<.01 I l l TABLE IV Analysis of Variance of Variable Error Source of Variance df MS F Between Subjects Learning (L) 1 .085378 11.47* Response (R) 1 .009631 1.29 Feedback (F) 2 .025985 3.49 LxR 1 .002045 <1 LxF 2 .004760 <1 RxF 2 .002335 <1 LxRxF 2 .011984 1.61 SwG 48 .007444 Within Subjects Blocks (B) 5 .001692 1.12 LxB 5 .001072 <1 RxB 5 .000924 <1 FxB 10 .003754 2.48 LxRxB 5 .001232 <1 LxFxB 10 .001299 <1 RxFxB 10 .002209 1.46 LxRxFxB 10 .000699 <1 SwGxB 240 .001515 *p<.01 112 TABLE V Analysis of Variance for the Simple Main Effect of Feedback At Each Learning-Response Combination in VE Source df MS F F at HL-•B 2 .002766 3.42 F at HL-•SP 2 .020238 24.99* F at LL-•B 2 .017008 2.18 F at HL-•SP 2 .004803 <1 Sw F at HL-•B 12 .000810 Sw F at HL-•SP 12 .002253 Sw F at LL-•B 12 .007798 Sw F at LL-SP 12 .018914 *p<.01 TABLE VI Analysis of Variance of Intra-Track Times Source of Variance df MS F Between S Learning (L) 1 1.838362 9.12** Feedback (F) 2 .263044 1.31 LxF 2 .444684 2.21 SwG 24 .201644 Within S Blocks (B) 49 .005149 <1 LxB 49 .003969 <1 FxB 98 .007785 1.23 LxFxB 98 .007642 1.21 SwGxB 1176 .006335 Times (T) 2 .043207 <1 LxT 2 .058886 <1 FxT 4 .2486043 2.70* LxFxT 4 .020852 <1 SwGxT 48 .091974 BxT 98 .004606 1.04 LxBxT 98 .004741 1.07 FxBxT 196 .004563 1.03 LxFxBxT 196 .004224 <1 SwGxBxT 2352 .004447 **p<.01 *p<.05 TABLE VII Analysis of Variance of SD of Intra-Track Times Source of Variance df MS F Learning (L) 1 .222269 5.38* Feedback (F) 2 .028012 <1 Blocks (B) 49 .001701 <1 LxF 2 .107257 2.59 LxB 49 .001333 <1 FxB 98 .002275 1.16 SwG 24 .014133 LxFxB 98 .002113 1.08 SwGxB 1176 .001966 *p<.05 115 APPENDIX C Tables o f Means 116 TABLE I The Effect of Type of Feedback and Level of Learning On a Self-Paced and Ballistic Response in Terms of CE Type of Response Level of Original Learning Type of Feedback in KR Withdrawal HF(sec) LF(sec.) NF(sec) HL .0083 -.0075 -.0139. Ballistic LL .0588 .0484 .2888 HL -.0088 -.0579 .1116 Self-Paced LL .0344 .1820 .1786 TABLE II The Blocks Effect at Each Level of Learning Level of Original Blocks Learning 1 2 3 4 5 6 HL LL .0110 .0029 .0098 .0041 .0042 -.0002 -.0007 .0958 .1471 .1661 .1840 .1987 117 TABLE III The Effect of Type of Feedback and Level of Learning On a Self-Paced and Ballistic Response in Terms of VE Type of Response Level of Original Type of Feedback in KR Withdrawal Learning HF (sec.) LF (sec.) NF(sec) High .0566 .0789 .0569 Ballistic Low .0854 .0837 .1261 High .0528 .0769 .1041 Self-Paced Low .0926 .1017 .1176 TABLE IV The Effect of Type of Feedback on the Intra-Track Times in an SP Response Type of Feedback Intra--Track Time 1 2 3 HF .1610 .1460 .1421 LF .1830 .1423 .1460 NF .1530 .1768 .1964 

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:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0077223/manifest

Comment

Related Items