Aviad Ozana¹ (), Yelda Alkan¹, James W Bisley^1,2; ¹Department of Neurobiology, David Geffen School of Medicine at UCLA, Los Angeles, California, USA, ²Department of Psychology,UCLA, Los Angeles, California, USA

The lateral intraparietal area (LIP) is thought to be a critical component of the neural network involved in spatial attention and eye movements, acting as a priority map. At the neural level, visual search could be implemented in the LIP priority map as parallel neural accumulators that indicate the location of a target when one of the accumulators reaches a threshold. To test this hypothesis, we adopted a covert target present/absent search paradigm using random dot patterns as stimuli (Alkan, Mirpour & Bisley, 2022). This allowed us to look at accumulation patterns of activity during search. We recorded the activity of single neurons in LIP while an animal performed the task. The animal was required to make a saccade towards a target and maintain fixation on target-absent trials. We varied the set size, coherence level, and had 100% valid attentional cues in some trials. Consistent with the parallel accumulator hypothesis, neural activity was correlated with the animal’s behavioral responses and task performance. Specifically, neural activity prior to both false alarm (target absent) and hit (target present) trials—where a saccade was eventually made to a stimulus—ramped up to similar peak response – a common threshold. Conversely, prior to miss and correct rejection trials, responses were significantly lower, presumably failing to reach the threshold. The level of activity preceding motion onset decreased with increased set size and was higher when attention had been cued. Lastly, we observed a different pattern of neural activity between correct and error responses. For example, compared to hits, false alarms were either preceded by elevated responses or had transient bursts and fluctuations before reaching threshold. Our findings are consistent with the hypothesis that covert visual search tasks use a parallel and a unified neural process aimed at target detection, in line with the signal detection model.