Bennett, Tom James; (2006) Temporal cognition as a feature of working memory. Masters thesis , UCL (University College London).

Preview

Text Bennett.Tom.J_thesis.pdf Download (15MB) | Preview

Abstract

Psychological studies of the way in which animals time intervals show a key scalar regularity. Namely, the standard deviation of the estimates of the interval length is proportional to the (typically nearly unbiased) mean of the estimates over multiple trials. This implies that the discriminative stimuli for animal behaviour in a timing task are subject to multiplicative noise. Despite a rich body of psychological models of timing, there is a dearth of physiologically-based accounts which explain this regularity. We propose a theory of interval timing based on the experimentally observed dynamical behaviour of cortical cells during Delayed Match to Sample working memory tasks. These neurons display a diverse array of repeatable temporal activity patterns (collectively termed persistent activity) in the delay period that follows the presentation of a stimulus in each trial. We treat these patterns as forming a temporal basis function representation of the time elapsed since the stimulus was shown. Recent electro-physiological data from parietal cortex, by Shaii, Bodner, Zhou and Fuster (2003) suggest that the standard deviation in the activity of individual cells at a given point in time across trials scales linearly the with mean activity of the cell at that time. Our model of sustained activity uses the inherent unreliability of synapses within a recurrent network of spiking cells to generate multiplicative noise internally. This mechanism, initially proposed by Shapiro, Wearden and Barone (2003), at all times renders the process of spike generation dependent on the post-stimulus history of synaptic releases. Persistent activity is generated without the need for bistable neurons by the percolation of spikes around the network. Activity levels are controlled by making the average probability that a neuron emits a spike, given that a spike has arrived at a presynaptic terminal, inversely proportional to the neuron's recurrent connectivity. Storage is initiated by a signal which resets the activity to a well denned and reproducible level. It is terminated either by an equivalent signal or as a consequence of the decay of the memory trace. Time estimates are made via threshold discrimination of the mean activity of the neural population. Random synaptic failures between the cells give rise to the proportional relationship between the standard deviation of the distribution of time estimates and the distribution mean over any set of trials. This makes our proposal consistent with relevant psychological and physiological results.

Download activity - last month

Export as JSONCSVXML

Download activity - last 12 months

Export as JSONCSVXML

Downloads by country - last 12 months

Export as JSONCSVXML

Archive Staff Only

View Item