UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A weighted summation framework for conjunctive predictions Yu, Ru Qi

Abstract

The mind readily learns predictive relationships in the environment where a cue predicts a specific outcome. This research examines a novel question: How does the mind spontaneously generate predictions when multiple cues associated with different outcomes are jointly presented. I propose a weighted summation framework to model human predictions: when encountering joint cues, the mind sums up the associated outcomes based on their respective probabilities. The conjunction that represents the overlap of the two outcomes would have the highest summed probability, and would be prioritized and consistently predicted. To examine the research question, Experiments 1-3 employed a spatial search paradigm. Participants were first exposed to cue-location associations where specific color (Experiments1-2) or texture cues (Experiment 3) predicted target appearance in specific locations. Then, two cues jointly appeared either side-by-side (Experiments 1-2) or into a new object (Experiment 3), and the target appeared in all locations with equal frequency. Following the two cues, there would be locations consistent with both of them, making these locations the conjunctive locations. The results showed that search time was faster when the target appeared in the conjunctive location. In Experiments 4-6, an attention tracking paradigm was used to extend the findings in Experiment 1. The results showed that when two cues jointly appeared, participants tended to first check the conjunctive location for the target, suggesting they made conjunctive predictions. Experiments 7-9 aimed to further extend the previous findings. Specifically, Experiment 7 replicated previous results with a conceptual paradigm, where each cue was associated with a conceptual category of objects (e.g., red predicts large objects). Experiments 8-9 examined weighted subtraction, where participants were exposed to associations between a pair of joint cues and specific spatial locations, and made predictions when they encounter a single cue from the pair. Experiments 10-11 examined the role of exposure in forming conjunctive predictions, where exposure was replaced with only explicit instructions (Experiment 10), or the strength of cue-outcome associations was reduced during exposure (Experiment 11). Overall, the results of the current research suggested that people tended to make conjunctive predictions when encountering joint cues, consistent with the weighted summation framework.

Item Media

Item Citations and Data

Rights

Attribution-NonCommercial-NoDerivatives 4.0 International