Noise correlations in the owl optic tectum are differentially impacted by unimodal vs crossmodal competitors

Multisensory integration in mammalian superior colliculus and its avian homolog, the optic tectum, is essential for rapid orienting towards salient stimuli. The benefits of combining across modalities is expected to depend in part on interneuronal noise correlations (NCs) and trial-to-trial response variance (Fano factor; FF), yet there is scant data on these topics. Here we used multielectrode arrays (MEAs) to record simultaneously from cohorts of single units in the deep layers of the owl optic tectum (OTd). Stimuli were presented individually or simultaneously as spatially aligned or non-aligned competitors. NC values varied from -0.35 to 0.94, an unexpectedly large range, and decreased only modestly as receptive fields diverged. Spatially aligned bimodal stimuli summed in a largely additive fashion over a wide range of response magnitudes. Most of the observed variance in bimodal NCs and FFs could be accounted for by an additive rule without accounting for internal noise. For non-aligned stimuli, crossmodal competitors decreased FFs, whereas unimodal competitors did not. Most striking, visual competitors decreased NCs for auditory drivers but increased NCs for visual drivers, indicating that the OTd network is differentially wired to process bimodal competitors. In total, these data provide novel descriptive information regarding the correlational structure of OTd neurons underlying multisensory processing. SIGNIFICANCE STATEMENTIn the owl auditory midbrain and its mammalian homologs, little is known regarding the ensemble response properties during multisensory integration. Using microelectrode arrays we found that noise correlations between pairs of neurons evoked by auditory, visual and bimodal stimuli were large and variable. When two non-aligned (competitor) stimuli were presented simultaneously, response reliability and noise correlations were driven down for crossmodal but not unimodal competitors, which suggests distinct processing strategies. This novel finding predicts that visual competitors drive an increase in localization accuracy for auditory targets.