Abstract
Jain, Fuller and Backus [PLoS ONE, 2010] reported that cue recruitment did not occur for three cues that were extrinsic to a rotating Necker cube. In other words, auditory and visual signals that were not conveyed by the object itself (so they were extrinsic to the cube) were not discovered and utilized by the visual system as cues that influenced the apparent rotation direction of the cube. However, research during the last year has shown that stronger learning can occur when training stimuli are not completely disambiguated by long-trusted cues, but rather are seen to have one conformation or the other as forced by priming with perceptual perseveration [Harrison and Backus, Vis Res, 2010; van Dam and Ernst, J Vis, 2010; Di Luca, Ernst, and Backus, Cur Biol, 2010]. We tested whether training with these “low-information” stimuli could cause an extrinsic signal to be recruited as a cue for apparent rotation direction of a 3D Necker cube. Cubes rotated about their vertical axis and were surrounded by an annular random-dot field rotating either clockwise or counter-clockwise in the plane of the display. The rotation directions of the random-dot field and Necker cube were correlated on training trials. At training trial onset, binocular disparity was present for 150 ms to control the percept for the entire trial (duration 1.5 s). Test trials contained the rotating field of dots but did not contain disparity, so cube rotation was ambiguous. Each session contained 96 training trials followed by four blocks (96 trials each) of mixed test and training trials in pseudo-random order. Under these new conditions, the random-dot field was recruited and utilized as a visual cue. The effect was small (58% seen as trained vs. 50% for chance). Thus, an extrinsic signal was learned when training stimuli had low information.
NIH R01 EY-013988, NSF BCS-0810944, HFSP RPG 3/2006.