Arbitration between cortical and sucortical behavior

When an automobile driver is at a red light, any visual change near the traffic light location might trigger initial movements preparatory to driving. But if visual discrimination of the shape, color and exact location of the change reveals it to be the wrong traffic signal, this information can suppress or cancel an inappropriate behavioral response. The premise of this study is to put subcortical vision (detection) and cortical vision (pattern discrimination) in direct conflict for the control of behavior in this manner, and determine how this conflict is resolved.

In rats, some visual tasks required primary visual cortex (V1), including image discrimination, whereas other visual tasks are V1-independent, including reporting stimulus location when only one stimulus is presented (Petruno et al., 2013). We hypothesize that V1-independent tasks are supported by the superior colliculus (SC).

When rats are trained to discriminate images invariant to the image size and rotation, this requires sometimes orienting to the less salient visual target. Rats perform this with high accuracy on average, but errors occur disproportionally in trials when the distractor is more salient. Our hypothesis is that these trials put V1 and SC in conflict or competition for control of behavior.

Behavioral evidence for competition between SC and V1. A. Rats were trained to respond at the behavioral response port near the statue (V1-independent). B. The same rats were trained to select the statue not the shuttle, invariant to independently varying sizes (V1-dependent). C. Performance of two rats after training to asymptotic performance. Color indicates Target size: small (red), medium (green) or large (blue). Horizontal lines indicate performance on Task A (Location). Symbols indicate performance on Task B (Salience Invariant); error bars show 95% confidence intervals. As distractor size (salience) increased, rats made an increasing number of error responses on the V1-dependent discrimination task (unpublished results).

We are now developing a similar task for mice, and using optogenetics, electrophysiology, and machine-learning-based pose estimation to observe the dynamics with which the competing visual targets vie for determination of motor control, in trials in which cortex wins versus loses the contest for control of behavior. This work will establish a new model system for dissecting the neural circuitry and computational algorithms through which cortical processing exerts influence over cortex-independent sensory-guided behaviors.

This simple case relates to broader questions about how conscious, intentional choices shape or override implicit or innate behavioral drives, and how factors like stress or fatigue affect the balance of power between cortex and subcortex.

This project is a collaboration between the Reinagel Lab at UCSD and the Scanziani Lab at UCSF and is supported by the National Science Foundation.