Efficient training protocol for rapid learning of the two-alternative forced-choice visual stimulus detection task

Shogo Soma,Naofumi Suematsu,Satoshi Shimegi

doi:10.14814/phy2.12060

Abstract

The potential of genetically engineered rodent models has accelerated demand for training procedures of behavioral tasks. Such training is generally time consuming and often shows large variability in learning speed between animals. To overcome these problems, we developed an efficient and stable training system for the two‐alternative forced‐choice (2AFC) visual stimulus detection task for freely behaving rodents. To facilitate the task learning, we introduced a spout‐lever as the operandum and a three‐step training program with four ingenuities: (1) a salient stimulus to draw passive attention, (2) a reward‐guaranteed trial to keep motivation, (3) a behavior‐corrective trial, and (4) switching from a reward‐guaranteed trial to a nonguaranteed one to correct behavioral patterns. Our new training system realizes 1‐week completion of the whole learning process, during which all rats were able to learn effortlessly the association between (1) lever‐manipulation and reward and (2) visual stimulus and reward in a step‐by‐step manner. Thus, our new system provides an effective and stable training method for the 2AFC visual stimulus detection task. This method should help accelerate the move toward research bridging the visual functions measured in behavioral tasks and the contributing specific neurons/networks that are genetically manipulated or optically controlled.

Highlights

IntroductionPigmented rodents have become a popular model in visual neurosciences (Laplante et al 2005; Goard and Dan 2009; Kang and Vaucher 2009; Kang et al 2013; Soma et al 2013a,b), because the specific neurons of rodents are relatively easy to genetically manipulate or optically control (Adesnik et al 2012; Olsen et al 2012; Nienborg et al 2013; Vaiceliunaite et al 2013)
These techniques have helped create a comprehensive network of visual information processing at the neuronal level. Scanziani and his colleagues discovered in mice that the gain of a visual response to grating stimuli in the primary visual cortex (V1) is controlled by layer 6 neurons (Olsen et al 2012) and that the spatial summation tuning properties of excitatory V1 neurons is formed by somatostatin-positive inhibitory neurons (Adesnik et al 2012)
By chewing on the tip of the spoutlever and pulling it up, rats could acquire the reward directly from the tip. The difficulty in this learning process is to awaken animals to the specific behavior associated with the reward delivery

Summary

Introduction

Pigmented rodents have become a popular model in visual neurosciences (Laplante et al 2005; Goard and Dan 2009; Kang and Vaucher 2009; Kang et al 2013; Soma et al 2013a,b), because the specific neurons of rodents are relatively easy to genetically manipulate or optically control (Adesnik et al 2012; Olsen et al 2012; Nienborg et al 2013; Vaiceliunaite et al 2013).

Methods

Results

Conclusion