Giarrocco, F. & Averbeck, B. B. (2026). Nature Communications 17:8443. doi:10.1038/s41467-026-75349-2.
Key takeaways
From the reading, NOTE-947.
- Task. Cue tone left or right (side blocked), delay of 1.0–1.8 s, target tone; release a bar within 700 ms on a match, keep holding on a non-match. Accuracy around 80%; match response times 564 and 601 ms.
- Sensory flow, one way. Cue side is decoded first in A1 (25 ms against 71–87 ms in PFC) and is predictable in PFC from A1’s past, never the reverse.
- Choice flow, both ways. Choice is decoded earlier in PFC; it becomes predictable in A1 from PFC 315–365 ms after target onset and in PFC from A1 about 40–55 ms later; both peak near 500–520 ms.
- Errors. Cue coding is intact on error trials; errors arise in the target comparison, and false match reports show first in A1.
- What it cannot say. “our paradigm does not allow us to dissociate decision-making from the motor response” (p. 9); “the correlational nature of our data precludes direct causal inference” (p. 9). The “editing” of sensory input by PFC feedback is Discussion, not result.
Why this is here
Registered at the owner’s request (U9) for its bearing on CLAIM-341: choice-related information sent from prefrontal cortex into primary sensory cortex and returned. The reading answers the two questions the registration left. The exchange is within the trial, and nothing in the paper tests whether it reshapes later sensory processing, so it is evidence of a synchronic exchange, which the claim sets aside, not of the diachronic form. And the response is motor, with the flow measured in the window of its preparation, so the paper cannot separate a judgement–observation loop from an action–perception one. It neither supports nor undermines the claim. Its value is a design: the same analysis on a signal measured before the next stimulus, with decision and report dissociated, would bear on the diachronic form.