The visual system is opportunistic, incorporating information from non-visual cues whenever possible to enhance its visuomotor control. This dramatically enhances its ability to detect and respond to environmental events and is evident in the response properties of deep layer superior colliculus (SC), a major midbrain visuomotor structure involved in detection, and orientation behavior. Using the visual-auditory SC neuron as a model, we showed that this process of visual multisensory integration must be instantiated during postnatal life based on experience with visual and non-visual (auditory) cues. Surprisingly, this experience is useful only in the presence of inputs from sensory-specific visual and auditory subdivisions of association cortex. In their absence SC neurons will develop the ability to respond to visual and auditory cues independently, but will not be able to integrate these cues to enhance their responses and the visuomotor behaviors that depend on them. But if given appropriate training later in life and in the presence of these cortical influences, they an still acquire this fundamental capability. These findings suggested the presence of substantial adult plasticity, a finding of more than passing interest to those with congenital visual or hearin deficits who have no opportunity to develop V-A multisensory integration until intervention corrects the underlying sensory defect. But, the practical use of this information is limited by poor understanding of the biological constraints and impact of age on this developmental process. We are now poised to determine how these cortical afferents accomplish this task, and what specific information they communicate to their SC target neurons. We posit that these sensory-specific cortical regions must form a functional ensemble with their target SC neuron for this developmental process to take place, and that rather than forming a simple associations, the ensemble encodes their relational statistics and uses this information to tune the properties of the SC neurons that will ultimately integrate this information. But, because of inherent features of the SC (e.g., its sensory topographies) the process is not veridical, but biased toward cross-modal cues in spatial and temporal concordance: those likely derived from the same event and, thus, possible targets of SC-mediated behaviors. Although this process is less rapid and efficient in adulthood, we hypothesize that this is because the circuit becomes less capable of extracting the relevant cross-modal statistics from the ambiguities that characterize normal environments, not because its fundamental mechanisms of plasticity have been compromised We propose to test each of these hypotheses in the current proposal.

R01EY024458

1986-08-01

2019-02-28