Effect of cues, context, and collective judgement on perceptual decision making

Saha Roy, Tiasha (2021) Effect of cues, context, and collective judgement on perceptual decision making. PhD thesis, Indian Institute of Science Education and Research Kolkata.

Text (PhD thesis of Tiasha Saha Roy (14RS073)) 14RS073.pdf - Submitted Version Restricted to Repository staff only Download (8MB)

Abstract

The aim of this thesis is to explore computationally the neural mechanisms underlying perceptual decision making while human subjects perform different complex real-world tasks. We at first provide empirical evidence of how individuals get manipulated by others’ decisions while performing a face/car identification task. We propose a two-component mixture model based on variants of the gamma distribution to explain the decisions taken by the participants under the influence of a decision cue. Participants were ranked based on the influence the cues elicited on their decision and the findings were subsequently corroborated by the subsequent neural analysis. Next, we used univariate and multivariate analyses of EEG signals to explore the influence of contextual guidance on visual search in natural scene images. We demonstrate that it is possible to decode contextual location from target absent images and the decoding accuracy was dependent on task difficulty. We also demonstrate that for easy target detection tasks, contextual guidance facilitation starts as early as 200 ms post stimulus onset whereas, for difficult tasks, the contextual effect was relevant at a later stage alluding to the role of task difficulty in mediating the neural timeline of contextual guidance. In the third project we address the question whether target detection error in the said visual search task could be minimised by aggregating individual decisions to form group decisions using well known pooling algorithms. We demonstrate that the classification error rates from individual judgement were found to increase with increasing task difficulty. This error could be significantly reduced upon combining the individual decisions using group aggregation rules. We also try to explore if group aggregation benefits depend on the correlation between the individual judgements of the group and our results seem to suggest that reduced inter-subject correlation can improve collective decision making for a fixed difficulty level.

Actions (login required)

View Item