Sometimes it is the lack of information, other times it is the abundance of information that hampers efficient decision making. In a rich environment, finding the relevant features from a flurry of irrelevant garbage is key to reaching the right conclusion. The brain needs to address this problem.
Even very simple experiments convincingly demonstrate that the brain is actually mastering this gating of information: in the attentional blindness paradigm one is tasked to count passes in a video showing a handful of people playing basketball. During the intense game a gorilla is passing through the players. Strangest than the presence of a gorilla in a basketball scene is the fact that the gorilla goes unnoticed at the overwhelming majority of viewers.
Is it the visual cortex which abandons on irrelevant information, or does this happen somewhere higher along the processing hierarchy? With our amazing collaborators at UCLA, the Golshani lab, we investigated mice during a task in which they were presented with both visual and auditory information and were required to make decisions on either the visual or the auditory modality. Since the modality that determined how reward could be obtained changed during the experiment, the relevance of information was actually not constant. This made it possible to assess how stimulus representation is affected by changing relevance. Reassuringly, the visual cortex did its job: the visual stimulus was represented irrespective of the relevance of this modality. However, in a region that is responsible for identifying the right behavioral strategy, the Anterior Cingulate Cortex (ACC) the irrelevant stimulus modality was systematically suppressed when the particular modality became irrelevant. Training a recurrent neural network on an identical task revealed a specific mechanism that leads to such suppression: contextual gating through mutual inhibition between neural subpopulations. A signature of this mechanism was identified in in silico experiments, which we could identify in our animal experiments as well.
The study was spearheaded by Marci Hajnal, and was greatly supported by ideas from Zsombor Szabó, as well as experiments by Andi Albert.
The paper is published at Nature Communications (https://doi.org/10.1038/s41467-024-49845-2).
Further reading: HFSP news