Event Details

After Pavlov (1927), it seemed clear that the visual system should be trainable by means of paired association (Fieandt, 1936; Hebb, 1949; Brunswik, 1953, 1956; Smedslund, 1955). But this prediction was not confirmed and today perceptual learning is often defined as an improvement in the ability to discriminate that comes with practice. Yet instances of associative learning have been documented for perceptual appearance and the modern view of perception as near-optimal inference requires this form of learning.  We now have a positive result in a direct test of this proposition, using "cue recruitment" experiments adapted from Pavlov's classical conditioning paradigm (Haijiang et al., 2006).    Trainees viewed movies of perceptually bistable Necker cube stimuli. On training trials, perceived rotation direction was forced by the addition of trusted cues (stereo and occlusion). Critically, arbitrary signals were also added, contingent on the trusted cues. On test trials stimuli contained the new signals but not the trusted cues.  The new signals often became cues: they acquired the ability to disambiguate the percept on their own.  We also report that the learning is incremental (cf. Gallistel, 2004) and that, unlike cases of "learning to learn" it slows down after counter-conditioning.  We aim to understand this learning by describing it in the frameworks provided by Bayesian inference and statistical machine learning.