Abstract
Cue recruitment studies (e.g. Haijang et al., 2006 PNAS) show that the visual system can be trained to disambiguate an ambiguous 3D rotation, such as the case of a movie of a rotating Necker (wire frame) cube. This acquired bias is specific to retinal location (Harrison & Backus, 2010 J Vis) and lasts at least four weeks. Resistance to reverse training is greater after viewing ambiguous stimuli without disparity, than disambiguated stimuli with disparity (Harrison & Backus, 2010 Vis Res; van Dam & Ernst, 2010 J Vis). This resistance was attributed to greater learning from “difficult to interpret” ambiguous stimuli. An alternative interpretation is that the ambiguous test stimuli used during reverse training resembled the ambiguous stimuli during initial training. Here we used stimuli that contained binocular disparity during reverse training, and the question was whether initial training would be more effective to block reverse learning. On Day 1, we measured participants' sensitivity to disparity using a staircase procedure. On Day 2, participants in Group A saw mostly non-disambiguated trials, the appearance of which was controlled by a few disambiguated trials (Harrison & Backus, 2010 Vis Res), and participants in Group B saw stimuli that contained disparity (1, 2, and 3 times threshold as measured on Day 1). On Day 3 both groups received reverse training on stimuli that contained disparity. If learning was specific to stimulus type, then resistance to retraining would be greater in Group B. If “difficult” or ambiguous stimuli simply cause stronger learning (or more resistant bias), then resistance to retraining would be greater in Group A. Results from four observers suggest that the second interpretation is correct, and that non-disambiguated stimuli do cause greater learning than disambiguated stimuli.