Abstract
The general goal of this work is to develop a visual search task that allows decomposition of functionally distinct processing stages through the analytic logic of separate modifiability. Monkeys are trained to search for an elongated color singleton among distractors. The orientation of the singleton specifies the stimulus-response mapping rule; pro-saccade for a vertical singleton and anti-saccade for a horizontal singleton. Distractors could be square or elongated. Elongated distractors create a new task demand; singleton and distractor orientation can be congruent (specifying saccades to same endpoint) or incongruent (saccades to opposite endpoints). Performance of this task requires multiple, sequential operations: localizing of the color singleton, encoding the shape of the singleton, encoding the instructed stimulus-response mapping, selecting the endpoint of the saccade, and preparing and initiating the saccade. Are errors in this task due to incorrect identification of the singleton, incorrect stimulus-response mapping, or both? Performance was impaired in anti-saccade (44% correct) versus pro-saccade (86%) trials. Errors revealed 3 patterns: (1) errant anti-saccades avoided the singleton (3% to singleton, 94% to distractors); (2) errant anti-saccades were concentrated on stimuli closer to the correct endpoint than to the singleton (set size 6, 64% to distractors flanking the correct endpoint, 28% to distractors flanking the singleton); and (3) errant saccades were primarily made to vertical stimuli (64 ± 11% to vertical, 13 ± 3% to square, 14 ± 5% to horizontal). These observations indicate that errors are not due to incorrect singleton localization or identification, but instead are due to either (a) incorrect stimulus-response mapping or (b) the use of an unintended strategy. These results will guide future studies of the neural correlates of the locus of failures of the sequential operations accomplishing this task.
Meeting abstract presented at VSS 2018