High order coordination of cortical spiking activity modulates perceptual accuracy

Research question and theoretical context

• The study probes whether temporally precise coordination of spiking within and between cortical populations influences perception beyond changes in firing rate.

• It engages a central debate: while firing rates reliably encode sensory content, rapid, high-order temporal coordination (beyond pairwise correlations) could modulate cortico-cortical signaling and perceptual accuracy, aligning with concepts of multi-layer synchronous spikes, coordination-based encoding, synchronous oscillations, and efficient driving of postsynaptic targets.

Methods

• Experimental design centers on simultaneous multiunit recordings from early and mid-level visual cortex (areas V1 and V4) in behaving macaques.

• 16-channel linear arrays were placed to maximize overlapping receptive fields across V1 and V4; up to 14 cells per area per session, totaling 293 significantly stimulus-modulated neurons across sessions.

• Monkeys performed a two-interval orientation discrimination task: two natural scenes were flashed (target and test); the test differed by a near threshold rotation (0° match or 3–5° non-match).

• Coordinated spiking was detected as near-coincident (5 ms) events among neuron ensembles; significance was assessed against jittered spike trains (±10 ms, 20 iterations) to isolate precise timing from slow rate fluctuations.

• Analyses included: (i) population decoding from firing rates with SVMs, (ii) decoding based on high-order coordination rates for pairs, triplets, and quartets, (iii) correlation of coordination with perceptual outcomes (correct/incorrect), and (iv) cross-area V1–V4 dynamics, including time-lag analyses and cross-correlograms. Simulations with matched firing statistics but without coordination served as controls.

Key findings

• High-order coordination (three or more cells) is time-locked to stimulus presentation and carries information about perceptual reports, independent of firing-rate changes.

• Decoding analyses showed: firing-rate-based decoders for orientation performed in both V1 and V4 on correct trials, but on incorrect trials only V1 decoding remained significant; V4 failed to distinguish stimuli in incorrect trials.

• In contrast, perceptual accuracy tracked high-order coordination in V4, but not in V1, and correlated with feedforward coordination between V1 and V4. Specifically, correct trials exhibited markedly higher high-order coordination in V4 (mean ~11.32 events, triplets ~3.7, quartets ~5.6) vs incorrect trials (mean ~2.31; triplets ~0.8, quartets ~1.5). V1 showed much smaller differences (≈0.62 vs ≈1.03).

• V1–V4 feedforward coordination peaked when V4 lagged V1 by ~+25 ms, indicating feedforward transmission, not feedback, underlies the perceptual effect. High-order cross-area events were statistically associated with behavior (p<0.02), whereas pairwise coordination was not.

• Ideal observer analyses indicated that V4 high-order coordination could partially separate correct vs incorrect trials (AROC ≈ 0.66 for a representative quartet; firing rates yielded ≈0.49). Across sessions, high-order coordination in V4 predicted perceptual accuracy more reliably than firing rates.

Interpretation and context

• Visual stimuli are represented by firing-rate modulations, but perceptual accuracy hinges on temporally precise spiking coordination within V4 and along feedforward V1→V4 pathways.

• The results support a framework where perception emerges from successive transformations across local and long-range networks, with high-order coordination serving as a mechanism to propagate information efficiently to downstream areas.

Theoretical implications

• The findings resonate with theories positing that coordinated spiking enhances inter-areal signaling and perceptual decisions, including multi-layer synchrony, structured spike words, and rapid, coordinated drive of postsynaptic targets.

• They emphasize a role for fast, feedforward coordination—not merely rate increases or feedback—that supports accurate perceptual reporting.

Limitations and future directions

• Limitation: inability to identify the exact neuronal identities comprising coordinated words; analyses focused on patterns without defining specific ensembles.

• Future work: extend to larger ensembles, link specific neuron groups to coordination patterns, and dissect the synaptic mechanisms that enable feedforward high-order coordination to boost perceptual accuracy.

Synthesis and takeaway

• The study advances understanding of neural coding by demonstrating that perceptual accuracy in vision is tied to temporally precise high-order coordination within and between cortical networks, especially in V4 and along feedforward V1→V4 pathways, beyond what firing rates alone convey. This reinforces a view of perception as emergent from coordinated population dynamics across hierarchical cortical circuits.