Sensory learning – the ability of our perceptual systems to exhibit change and improvement in response to sensory input – is an important class of processes that allows perception to adapt meaningfully to the environment. It can include improvements based on repeated exposure to stimuli, called perceptual learning (PL), and learning patterns governing stimuli, called statistical learning (SL). Efficiently combining information from multiple modalities, called multisensory integration (MI), has also been shown to interact with these processes. In a series of experiments, we investigate whether PL, SL, and MI might each comprise multiple underlying mechanisms and whether those mechanisms might intersect, using behavioral and neurological methodology.

In the first chapter, we conduct three experiments using an SL paradigm to show that SL may comprise multiple processes. Experiment 1 demonstrates our ability to find a dissociation between behavioral outcome measures; Experiment 2 replicates Experiment 1 using a different task; and Experiment 3 replicates it further while investigating whether the learning is associative or representational. In the second chapter, we analyze electroencephalogram (EEG) alpha power as a proxy for attentional resources during PL training. Behavioral results suggest that PL occurred and electrophysiological results suggest that after training, participants were able to perform the task using fewer resources and to allocate those resources more efficiently. This latter result did not apply specifically to trained stimuli, suggesting that multiple mechanisms may be at work during PL. The third chapter uses a novel training paradigm that combines PL, SL, and MI to investigate the extent to which these processes share common mechanisms and how their interactions impact perception. Across two sets of experiments, participants performed a discrimination task on audio-visual stimuli that appeared according to controlled spatio-temporal statistics. Behavioral data show some PL, SL, and MI effects, and EEG data hint at an SL and PL relationship with alpha power. The results of this study, although not all statistically significant, provide insight into the underlying mechanisms of PL, SL, MI.

Altogether, we provide evidence that sensory learning and other perceptual processes should not be treated as unitary mechanisms but instead should be investigated in terms of their manifold natures.