Keywords
pyramidal cell, binding, attention, feedback, biased competition
Abstract
Binding of visual information is crucial for several perceptual tasks. To incrementally group an object, elements in a space-feature neighborhood need to be bound together starting from an attended location (Roelfsema, TICS, 2005). To perform visual search, candidate locations and cued features must be evaluated conjunctively to retrieve a target (Treisman&Gormican, Psychol Rev, 1988). Despite different requirements on binding, both tasks are solved by the same neural substrate. In a model of perceptual decision-making, we give a mechanistic explanation for how this can be achieved. The architecture consists of a visual cortex module and a higher-order thalamic module. While the cortical module extracts stimulus features across a hierarchy of spatial scales, the thalamic module provides a purely spatial relevance map. Both modules interact bidirectionally to enter locations of task-relevance into thalamus, while allowing integration of context with local features within cortical maps. This integration realizes the model's binding mechanism. It is implemented by pyramidal neurons with dynamical basal and apical compartments performing coincidence detection (Larkum, TINS, 2013). The basal compartment is driven by bottom-up feature information and produces the neuron's output, the apical compartment computes top-down contextual information akin to an interaction skeleton (Roelfsema&Singer, Cereb Cortex, 1998). Apical-basal integration yields an up-modulation in neuron activity binding it to the attended configuration. Gating information from thalamus restricts this integration to task-relevant locations (Saalman&Kastner, Curr Op Neurobiol, 2009). By model simulations, we show how altering the apical compartment's operation regime steers binding to either perform search or incremental grouping.
Included in
Artificial Intelligence and Robotics Commons, Cognition and Perception Commons, Cognitive Neuroscience Commons, Computational Neuroscience Commons
A Dynamical Model of Binding in Visual Cortex During Incremental Grouping and Search
Binding of visual information is crucial for several perceptual tasks. To incrementally group an object, elements in a space-feature neighborhood need to be bound together starting from an attended location (Roelfsema, TICS, 2005). To perform visual search, candidate locations and cued features must be evaluated conjunctively to retrieve a target (Treisman&Gormican, Psychol Rev, 1988). Despite different requirements on binding, both tasks are solved by the same neural substrate. In a model of perceptual decision-making, we give a mechanistic explanation for how this can be achieved. The architecture consists of a visual cortex module and a higher-order thalamic module. While the cortical module extracts stimulus features across a hierarchy of spatial scales, the thalamic module provides a purely spatial relevance map. Both modules interact bidirectionally to enter locations of task-relevance into thalamus, while allowing integration of context with local features within cortical maps. This integration realizes the model's binding mechanism. It is implemented by pyramidal neurons with dynamical basal and apical compartments performing coincidence detection (Larkum, TINS, 2013). The basal compartment is driven by bottom-up feature information and produces the neuron's output, the apical compartment computes top-down contextual information akin to an interaction skeleton (Roelfsema&Singer, Cereb Cortex, 1998). Apical-basal integration yields an up-modulation in neuron activity binding it to the attended configuration. Gating information from thalamus restricts this integration to task-relevant locations (Saalman&Kastner, Curr Op Neurobiol, 2009). By model simulations, we show how altering the apical compartment's operation regime steers binding to either perform search or incremental grouping.