Back

Focusing perceptual attention in one modality constrains subsequent learning in another modality

Vasilev, D.; Watanabe, M.; Logothetis, N. K.; Totah, N. K.

2022-01-23 neuroscience
10.1101/2022.01.22.477334 bioRxiv
Show abstract

Contemporary neuroscience and psychiatry suggest that attention to decision outcomes guides rule learning by adjusting stimulus-outcome associations. Separately, sensory neurophysiology conceptualizes attention as a filter that improves perception. Here, we show that the contemporary view is incomplete by demonstrating an unconventional and novel effect of perceptual attention on subsequent outcome-based rule learning. Moreover, we show for the first time in rodents that, like in primates, this attentional process involves tuning of modality specific cortico-cortical interactions. We designed a novel head-fixed rat-on-a-treadmill apparatus and used it to train rats to discriminate auditory-visual stimuli using one modality and then reduced stimulus discriminability in that modality. We observed perceptual learning suggesting engagement of perceptual attention. Moreover, engaging visual perceptual attention resulted in more saccades and increased frontal-visual cortex EEG Granger causality relative to engaging auditory perceptual attention. We then presented novel and easily discriminable stimuli in both modalities and measured outcome-driven learning in the other modality. Learning was slower after engaging perceptual attention. Our work suggests that a more complete description of learning requires integrating these previously siloed concepts of attention. Moreover, treating impaired set-shifting as a trans-diagnostic symptom may require targeting different neural circuits for perceptual attention or outcome-based attention depending on which type of attention is impaired in each neuro-psychiatric disorder.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.