Behavioral optogenetics in nonhuman primates; a psychological perspective

Abstract

Optogenetics has been a promising and developing technology in systems neuroscience throughout the past decade. It has been difficult though to reliably establish the potential behavioral effects of optogenetic perturbation of the neural activity in nonhuman primates. This poses a challenge on the future of optogenetics in humans as the concepts and technology need to be developed in nonhuman primates first. Here, I briefly summarize the viable approaches taken to improve nonhuman primate behavioral optogenetics, then focus on one approach: improvements in the measurement of behavior. I bring examples from visual behavior and show how the choice of method of measurement might conceal large behavioral effects. I will then discuss the "cortical perturbation detection" task in detail as an example of a sensitive task that can record the behavioral effects of optogenetic cortical stimulation with high fidelity. Finally, encouraged by the rich scientific landscape ahead of behavioral optogenetics, I invite technology developers to improve the chronically implantable devices designed for simultaneous neural recording and optogenetic intervention in nonhuman primates.

Graphical abstract

Fig. 1

Projective field. a) The activity of a neuron anywhere in the sensory cortex can be perturbed by a subset of external stimuli. Parts of the sensory space that influence the responses of the neuron are defined here as the “receptive field”. Note that the sensory space includes non-spatial aspect of the stimulus as well. Similarly, perturbation of the activity of a neuron can induce various behavioral changes. The subset of all psychophysically measurable behaviors that are altered following perturbation of a neuron is referred to as the “projective field” here. b) All tasks are not equally affected by perturbation of a given neuron. The behavioral alteration vector casts shadow projections of various sizes on different measurement axes. Conversely, the projective vector can be constructed, with behavioral measurement on multiple axes.

Fig. 2

The Cortical Perturbation Detection (CPD) task. a) The animal fixates on the screen, then an image is shown. A locus in the inferior temporal cortex is optogenetically stimulated randomly in half of the trials. The animal is rewarded for detection of this cortical stimulation impulse by making a saccade to one of the two consequently presented targets. We noticed that the performance of the animals in CPD highly depends on the properties of the image on the screen. This is because the animal detects a “visual perceptual event” to perform the task. This mental perceptual events interacts with the image and can be captured and described by perturbing the image on various visual dimensions. Using machine learning it is possible to find the one visual perturbation that maximizes the behavioral false alarm rate. We call such an image “perceptogram” (see text) as looking at it makes the animal think its brain is stimulated in the absence of stimulation. b) While the CPD task is very sensitive, it lacks specificity. Nevertheless, given the wide dynamic range provided by the large CPD effect sizes, also because of dependence of the task on the viewed images, it is possible to gain the lacking specificity by varying the visual stimulus across multiple image dimensions. This allows profiling of the projective vector over many possible tasks and dimensions of interest. A “perceptogram” is an example of projective field profiling in the image domain, but the basic logic holds for other domains, modalities and cortical regions.