Salience by competitive and recurrent interactions: Bridging neural spiking and computation in visual attention.

Abstract

Decisions about where to move the eyes depend on neurons in frontal eye field (FEF). Movement neurons in FEF accumulate salience evidence derived from FEF visual neurons to select the location of a saccade target among distractors. How visual neurons achieve this salience representation is unknown. We present a neuro-computational model of target selection called salience by competitive and recurrent interactions (SCRI), based on the competitive interaction model of attentional selection and decision-making (Smith & Sewell, 2013). SCRI selects targets by synthesizing localization and identification information to yield a dynamically evolving representation of salience across the visual field. SCRI accounts for neural spiking of individual FEF visual neurons, explaining idiosyncratic differences in neural dynamics with specific parameters. Many visual neurons resolve the competition between search items through feedforward inhibition between signals representing different search items, some also require lateral inhibition, and many act as recurrent gates to modulate the incoming flow of information about stimulus identity. SCRI was tested further by using simulated spiking representations of visual salience as input to the gated accumulator model of FEF movement neurons (Purcell et al., 2010, 2012). Predicted saccade response times fit those observed for search arrays of different set sizes and different target-distractor similarities, and accumulator trajectories replicated movement neuron discharge rates. These findings offer new insights into visual decision-making through converging neuro-computational constraints and provide a novel computational account of the diversity of FEF visual neurons.

Diffusion theory of the antipodal “shadow” mode in continuous-outcome, coherent-motion decisions.

“…In contrast, our theoretical inclination is to view the processes that compute drift rates as inherently nonlinear—plausibly, as following a MAX, or maximum-of-outputs, rule (e.g., Eckstein et al, 2000) in which the response depends on the largest or the strongest of a set of competing alternatives. MAX-rule dynamics arise naturally in competitive interaction neural architectures, like the competitive interaction model of attention and decision-making of Smith and Sewell (2013) and the related neural model of Cox et al (2022), in which the excitation or activation in one process inhibits all of the other processes in its neighborhood. When these kinds of competitive interactions are controlled by nonlinear dynamics like those analyzed by Grossberg (1980), then MAX-rule behavior occurs: The most strongly activated unit saturates and the activity of its competitors is suppressed.…”

“Reliable organisms from unreliable components” revisited: the linear drift, linear infinitesimal variance model of decision making

“…They found that RTs were better described by a model, which they called a "gated accumulator model," which had competitive interactions among accumulators, than they were by a model without such interactions. Cox, Palmeri, Logan, Smith, and Schall (2022) combined the competitive interaction model of attentional selection of Smith and Sewell (2013) with the gated accumulator model of Purcell et al to jointly model firing rates in frontal eye field neurons and distributions of RT in a saccade-to-target decision task. A novel feature of Cox et al's approach was that they were able to characterize the importance of different computational mechanisms, such as recurrence, feedforward inhibition, and lateral inhibition, to the firing rates of the individual neurons in their sample.…”

“…The evidence for lateral inhibition in search is unsurprising because successful search requires resolution of the competition between targets and distractors and lateral inhibition is an effective way to accomplish this, especially when it is nonlinear (Cox et al, 2022;Grossberg, 1980;Smith & Sewell, 2013;. However, an identified role for lateral inhibition in search leaves open the question of its role in tasks not involving inhibition or suppression of distractor stimuli.…”

Diffusion Theory of the Antipodal "Shadow'" Mode in Continuous-Outcome, Coherent-Motion Decisions

“…Eckstein et al, 2000) in which the response depends on the largest or the strongest of a set of competing alternatives. MAX-rule dynamics arise naturally in competitive interaction neural architectures, like the competitive interaction model of attention and decision-making of Smith and Sewell (2013) and the related neural model of Cox et al (2022), in which the excitation or activation in one process inhibits all of the other processes in its neighborhood. When these kinds of competitive interactions are controlled by nonlinear dynamics like those analyzed by Grossberg and colleagues (Grossberg, 1980), then MAX-rule behavior occurs: The most strongly activated unit saturates and the activity of its competitors is suppressed.…”