MAOZ SHAMIR
Projects
STDP and the transfer of sensory signals in the brain
Synaptic plasticity and the transmission of sensory information
In a recent project we studied the interplay between the transfer of sensory signal downstream the information processing pathway and activity dependent plasticity using the framework of the whisker system. Rats and mice move their whiskers back and forth rhythmically to scan their environment. The azimuthal location of a whisker can be inferred from the activity of whisking neurons that fire selectively to the phase along the whisking cycle. Preferred phases of whisking neurons have been found to be widely distributed. However, simple models for the transmission of the whisking signal predict a distribution of that is considerably narrower than empirically observed. We suggested that activity dependent plasticity in the form of spike-timing-dependent plasticity (STDP) may provide a solution to this conundrum.
In a modeling study we investigated STDP dynamics of the feed-forward connection along the whisking pathway. We found that for a wide range of parameters, STDP dynamics do not relax to a fixed-point solution. Consequently, the preferred phases of downstream neurons drift in time at a non-uniform velocity, which in turn induces a non-uniform distribution of the preferred phases of the downstream population. Thus, surprisingly, the transfer of sensory signal, response selectivity and the distribution of preferred stimuli evolve – not despite, but rather due to constant synaptic remodeling. This suggests that synaptic motility may not be a BUG but a FEATURE. Furthermore, our investigation provided several key empirical predictions to test this hypothesis.
Rhythmogenesis via STDP
Brain rhythms are widely believed to reflect numerous cognitive processes. Changes in rhythmicity have been associated with pathological states. Various experimental and theoretical studies have investigated rhythmic activity. The theoretical efforts have mainly been focused on neuronal dynamics, under the assumption that network connectivity satisfies certain fine-tuning conditions required to generate the desired rhythmic activity. However, it remains unclear how this fine-tuning is achieved. In recent studies we demonstrated how spike timing-dependent plasticity (STDP) can provide the underlying mechanism for tuning synaptic connectivity to generate rhythmic activity.
Can 'winner-take-all' underlie the computation of saliency?
The human brain is constantly met with a plethora of visual information. To manage this influx, it employs strategies to filter and prioritize specific segments of the visual scene, leading to selective attention. While significant research has been dedicated to understanding the behavioral and physiological aspects of selective attention, there remains a gap in bridging these two domains. A prevailing theory in the field posits that a winner-take-all competition among contextually modulated cells forms a saliency map, directing attention. However, this theory has not been critically examined, especially in the context of bridging physiological data with behavioral outcomes.
We employed the pop-out task paradigm to assess the accuracy of the winner-take-all mechanism in detecting salient objects. Our approach has been to investigate the accuracy of the winner-take -all readout and compare it to the empirically observed high accuracy in the task. Specifically, we focused on two mechanisms: the single-best-cell winner-take-all algorithm and a generalized population winner-take-all algorithm.
Our study revealed that neither algorithm could replicate the high accuracy observed in behavioral experiments. Specifically: The single-best-cell algorithm was significantly impacted by neuronal heterogeneity. The generalized algorithm, while robust to heterogeneity, was affected by observed neuronal noise correlations.
Thus, our results challenge the widely accepted winner-take-all theory, especially in the early stages of visual search. This underscores the need for a deeper understanding of the mechanisms guiding visual attention.
