Back

Intrinsic interval timing, not temporal prediction, underlies ramping dynamics in visual and parietal cortex

Huang, Y.; Shamsnia, A.; Chen, M.; Wu, S.; Stamm, T.; Medico, S.; Najafi, F.

2025-09-11 neuroscience
10.1101/2025.09.11.673960 bioRxiv
Show abstract

Neural activity following regular sensory events can reflect either the elapsed time since the previous event (temporal signaling) or temporal predictions and prediction errors about the next event (temporal predictive processing). These mechanisms are often confounded, yet dissociating them is essential for understanding neural circuit computations. We addressed this by performing two-photon calcium imaging from distinct cell types (excitatory, VIP, and SST) in layer 2/3 of the medial visual cortex (VIS) and the posterior parietal cortex (PPC) while awake mice passively viewed audio-visual stimuli under temporal contexts with different inter-stimulus interval (ISI) distributions. Computational modeling revealed distinct functional clusters of neurons, stimulus-activated (ramp-down) and stimulus-inhibited (ramp-up), with characteristic kinetics and area- and cell-type-specific biases. Importantly, all functional clusters were invariant to temporal predictability, shifted immediately when temporal statistics changed, and existed even in naive mice--findings inconsistent with predictions from temporal predictive processing frameworks. Elapsed-interval information could nonetheless be decoded, and heterogeneous kinetics underlay how well each cluster represented interval duration, leading to a distributed, learning-independent population code for time. These results demonstrate that activity in VIS and PPC arises from stimulus-evoked dynamics whose intrinsic evolution encodes elapsed time, rather than from temporal predictive processing, redefining the mechanistic origin of ramping and omission-related activity in visual and parietal cortices. We propose that temporal predictive processing may instead be implemented in other circuits or recruited in VIS/PPC during task-engaged behavior. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=157 HEIGHT=200 SRC="FIGDIR/small/673960v3_ufig1.gif" ALT="Figure 1"> View larger version (47K): org.highwire.dtl.DTLVardef@d560d2org.highwire.dtl.DTLVardef@c78193org.highwire.dtl.DTLVardef@be072org.highwire.dtl.DTLVardef@1bc9d04_HPS_FORMAT_FIGEXP M_FIG C_FIG

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.