Talk abstracts > Vezha Boboeva (Imperial College London)

Go back to Homepage.

Vezha Boboeva - Sainsbury Wellcome Centre

Thursday, June 1^st

Talk Session 1: How artificial neural networks make decisions

11h - 11h30 

Modelling memory effects in perceptual decision-making

Results from many decades of perceptual decision-making tasks are rife with a multitude of bias effects. In this talk, I will present a recent modelling work on the the central tendency bias, or contraction bias, where the judgment of the magnitude of items held in working memory appears to be biased towards the average of past observations. Contraction bias is assumed to be an optimal strategy by the brain, and commonly thought of as an expression of the brain's ability to learn the long-term statistical structure of sensory input. On the other hand, recency biases such as serial dependence are also commonly observed, and are thought to reflect the content of working memory.

Building on previous results from an auditory delayed comparison task in rats, I will propose a mechanistic model of the circuit that may be involved in generating such biases. In particular, I will show that a volatile working memory content susceptible to shifting to the past sensory experience - producing short-term sensory history biases - naturally leads to contraction bias. The network behaviour is consistent with a broad set of empirical findings and provide predictions of performance across different stimulus distributions and timings, delay intervals, as well as neuronal dynamics in putative working memory areas. Finally, I will show results from a set of human psychophysics experiments that validate our model.