Steinmetz and colleagues designed an experiment: A mouse in a chamber, focused on three screens. Images with varying contrasts appear. The mouse's task is to distinguish between these visual cues and make a choice.They used Neuropixels probes to record from ~30,000 neurons across 42 brain regions. Visual stimuli varied in contrast and could appear on either side, both, or neither. They are expected to turn the wheel towards the higher contrast, & should resist moving for no stimuli, which is rewarded. This two-alternative forced choice and Go/NoGo task tracks decision-making. Random rewards were given for equal contrast on both sides. Their study revealed distinct activity patterns and spatial distribution of neurons participating in action, choice, and engagement, enhancing decision-making and motor actions. They found that neurons in nearly all regions responded non-specifically during action initiation.
Building on Steinmetz's research, we investigated the neural basis of decision-making in Cori, an 11-12-week-old female mouse, over three sessions. By analyzing Local Field Potentials (LFPs) across multiple brain regions during different task epochs, we explored how neural activity informs behavior, with a focus on variations in visual contrast and behavioral choices.
Here we aim to answer how brain areas coordinate to transform visual input into action by hypothesizing that functional connectivity between visual, decision-making, and motor areas increases from stimulus onset to response execution, with a more pronounced increase in Go trials and high-contrast conditions.
Comprehensive view of the average LFP (Local Field Potential) activity across multiple brain areas and sessions for positive feedback conditions. During stimulus processing (~500ms), positive feedback trials show a pronounced and synchronized peak, especially in ACA and CA1, indicating more efficient processing of high-contrast stimuli. In the decision-making phase (500-1000ms), there's a large negative deflection followed by recovery, with more synchronized activity. Post-decision activity (after 1200ms) shows more aligned activity, suggesting better integration of decision outcomes. Brain areas like ACA, CA1, and SUB display larger amplitude changes in the positive feedback condition, reflecting stronger engagement in decision-making and memory processes. Visual areas (VISl, VISp, VISpm) exhibit larger amplitudes, indicating enhanced visual processing for high-contrast stimuli.
The heatmap shows LFP activity consistency across sessions 1 and session 3. VISp has a high positive correlation (0.86),indicating stable visual processing. DG dentate gyrus has a slight negative correlation (-0.15), suggesting variability in memory processes. Postsubiculum and CA1 show moderate positive correlations (0.68 and 0.64), indicating consistent cognitive and memory-related processes. This highlights stable visual and memory processing in POST and CA1, with more variability in DG Overall, positive feedback boosts coordination and amplitude in primary visual area VISp and DG, while negative feedback reduces and varies activity. This suggests VISp maintains consistency across sessions, whereas DG varies more with feedback type.