Abstract
Perceptual decision making tasks have a long history in the visual sciences to understand the organization of visual processing and to develop sophisticated models of response times in stochastic accumulator model frameworks. However, these tasks are often limited to one dimension of manipulation and the corresponding computational models are limited to one operation or stage of processing at which experimental manipulations are manifest, whereas in natural behavior decisions require the integration of information of different kinds from multiple sources. Integration of multiple sources is achieved through some processing architecture. The concept of processing architecture includes whether sources are processed in series or in parallel as well as whether or not all processes must be completed before a response is produced. Because current models do not require integration, neither the processing architecture used in a task nor the neural implementation of that processing architecture are known. To this end, we developed a multidimensional perceptual decision making task (Lowe et al., 2019). Two monkeys were trained on a GO/NO-GO visual search task, which involved explicit factorial manipulations of search difficulty and rule encoding difficulty. Performance was analyzed using systems factorial technology (Townsend & Nozawa, 1995) to assess the processing architectures used by each monkey. Here, we report the activity of frontal eye field neurons recorded during this task. We identified one sub-population of neurons modulated by search difficulty but not stimulus-response rule discrimination difficulty. We also identified a separate sub-population of neurons modulated by both factorial manipulations. These two sub-populations are key predictions of the processing architectures used by the monkeys assessed through pure behavior. The results confirm the utility of systems factorial technology in inferring the processing underlying behavior and provide first-of-its-kind insights into the neural substrates of multidimensional perceptual decision making.