The priority map hypothesis posits that visual attention is allocated according to the priority of objects in the visual scene. We and others have suggested that activity in priority maps during visual search might explain behavior based on the accumulation of evidence at each stimulus location (Purcell et al, 2012; Arcizet et al, 2018; Servant et al, 2019), however these studies have utilized simple stimuli that require little temporal integration of information. In this study, we use random dot stimuli in visual search to quantify how the accumulation of evidence changes under distinctive conditions within visual search. Two adult animals were trained on a visual search task in which 1, 2 or 4 random moving dot stimuli were presented. One subject indicated the presence of a target (a stimulus with rightwards motion) by making a saccade to the target. The other subject indicated the presence of a target by releasing one of two bars. The number of stimuli and the coherence of the stimuli were changed on each trial, such that all stimuli had the same coherence. Targets were present on 50% of trials for set size 1, 66% of trials for set size 2, 80% of trials for set size 4 in the saccade task and 66% of trials for set size 4 in the bar release task. In a subset of trials with a set size of 4, we also had a spatial cue condition and a feature-based cue condition. We found a decrease in performance, hit rate, and correct rejections for the greater set sizes. In the spatial cue and feature-based cue conditions, we found an increase in hit rates and decrease in reaction times. These behavioral data show that our task should allow us to measure the neural accumulation of evidence in a visual search task.