PSTH: Standard vs Omission Trials (V1 Local E/I Populations)

[anoushkac08]

Following up on the depth-resolved E/I coupling analysis. This is a plot on the differences in how excitatory and inhibitory populations respond to expected (standard) vs unexpectedly omitted (sequence_omission) stimuli in the Sequence Mismatch Block.
Methods: PSTH computed by aligning spikes to trial onset (-0.1s to 0.4s window, 10ms bins), averaged across all local SUA units at depth 2200µm (103E, 21I) classified by waveform width.
Result: Excitatory neurons show elevated firing following omission trials relative to standard trials, while inhibitory neurons show the opposite pattern, reduced firing following omissions. This is consistent with a prediction-error-like E/I decoupling.

I also attempted to compare this PSTH between high-coupling (400µm) and low-coupling (1000µm) depth zones, but found the result too noisy to interpret confidently given smaller per-bin sample sizes (23-42 neurons). Happy to share that analysis too if useful, but wanted to flag it as inconclusive rather than a clean finding.

[jeromelecoq]

I am assuming this is the average of all units? It might be worth looking into the diversity of units across both types of trials.
Units in superficial layers of the cortex will drastically have different behavior. I see inhibitory units as controlling the input gain through E/I balance. During omission, you are likely in a situation where the inhibitory network readjust its entire state to redirect the flow of information. So I would predict different inhibitory cells at different depth would have drastically different behaviors. The Layer II/III inhibitory cells vs layer 4/5 for example would have opposite dynamics during omission in that hypothesis