A Probabilistic Strategy for Understanding Action Selection

Abstract

Brain regions involved in transforming sensory signals into movement commands are the likely sites where decisions are formed. Once formed, a decision must be read out from the activity of populations of neurons to produce a choice of action. How this occurs remains unresolved. We recorded from four superior colliculus neurons simultaneously while monkeys performed a target selection task. We implemented three models to gain insight into the computational principles underlying population coding of action selection. We compared the population vector average (PVA)/optimal linear estimator (OLE) and winner-takes-all (WTA) models and a Bayesian model, maximum a posteriori estimate (MAP), to determine which predicted choices most often. The probabilistic model predicted more trials correctly than both the WTA and the PVA. The MAP model predicted 81.88%, whereas WTA predicted 71.11% and PVA/OLE predicted the least number of trials at 55.71 and 69.47%. Recovering MAP estimates using simulated, nonuniform priors that correlated with monkeys' choice performance, improved the accuracy of the model by 2.88%. A dynamic analysis revealed that the MAP estimate evolved over time and the posterior probability of the saccade choice reached a maximum at the time of the saccade. MAP estimates also scaled with choice performance accuracy. Although there was overlap in the prediction abilities of all the models, we conclude that movement choice from populations of neurons may be best understood by considering frameworks based on probability.

Deficits in reach target selection during inactivation of the midbrain superior colliculus

“…Activity in such a map does not encode the specific object features (e.g., color, shape) of potential movement goals; rather, it represents the physical salience and behavioral relevance of the goals. Target selection has been hypothesized to involve a competition among the goals in the priority map, leading to the selection of a single movement goal (1)(2)(3)(4)(5). Current evidence suggests that priority maps for target selection are distributed across a number of brain regions (6-10), and oculomotor areas, such as the SC, have been shown to possess the characteristics of a priority map for eyemovement tasks (7,(11)(12)(13)(14)(15)(16)(17).…”

“…Target selection is usually conceived of as a competition among potential movement goals (1)(2)(3)(4)(5), occurring in a priority map that encodes both the physical salience and behavioral relevance of each goal (6)(7)(8)(9)(10).…”

Blinks slow memory-guided saccades

“…Three monkeys (Macaca mulatta) were implanted with eye loops for monitoring eye movements and cranial cylinders for electrophysiological recording of single neurons, as described previously (Kim and Basso 2010;Li and Basso 2011). These animals participated in electrophysiological experiments ongoing in the laboratory.…”

Reach target selection in humans using ambiguous decision cues containing variable amounts of conflicting sensory evidence supporting each target choice

“…It is also reminiscent of the neural mechanisms underlying saccade and reach target selection in visual search tasks (Kim and Basso 2010;Purcell et al 2012;Song and McPeek 2010). Ultimately, neural recordings will be needed to provide more definitive evidence for the neural mechanisms leading to reach target selection in the CG task.…”

“…This has been extensively studied in the oculomotor system in tasks in which subjects chose the target for saccadic eye movements from among multiple distractors in visual-search paradigms or on the basis of the direction of variable-strength coherent visual motion in random-dot kinematogram (RDK) stimuli (Carpenter and Williams 1998;Churchland et al 2008;Ditterich 2006aDitterich , 2006bDitterich , 2010Hanes and Schall 1996;Horwitz et al 2004;Kim and Basso 2010;Kim and Shadlen 1999;Mazurek et al 2003;Palmer et al 2005;Purcell et al 2012;Ratcliff et al 2003Ratcliff et al , 2007Roitman and Shadlen 2002;Sato and Schall 2003;Shadlen et al 1996). These studies suggest that saccade target selection is accomplished by parallel neural circuits that continuously acquire sensory evidence for and against each target choice across time until the accumulated evidence in one circuit exceeds a decision criterion threshold.…”