Search bioRxiv⌕ Search

Biology subjects

Poggi, P. G.

Publications and source records attributed to Poggi, P. G..

2 recordsLinked to original sources

V1 interlaminar coherence decreases with interocular conflict

Resolving conflicting input from the two eyes is a fundamental challenge for the visual system. In the primary visual cortex (V1), such interocular conflict induces modest suppression of single neuron spiking, but the accompanying population-level dynamics remain poorly understood. Here we examined laminar multi-unit activity and interlaminar local field potential (LFP) coherence in macaque V1 during dichoptic stimulation and binocular rivalry flash suppression (BRFS). From laminar microelectrode recordings, we found that interocular conflict reliably reduces interlaminar coherence, particularly between granular and infragranular layers, suggesting altered temporal coordination across the cortical column. Strikingly, during BRFS, coherence remained reduced even when firing rates were unchanged. Moreover, interlaminar coherence is higher for perceptually dominant BRFS stimuli, indicating that coherence across V1 layers covaries with perceptual outcome in the absence of significant firing-rate differences. These findings show that the temporal dynamics of population coherence are a more stable signal of interocular conflict than spike rate modulation. SIGNIFICANCE STATEMENTThese findings suggest that V1 processes interocular conflict not only through modest rate changes but also through temporal coordination of population activity across cortical layers. Interlaminar coherence therefore offers a complementary perspective on V1s role during binocular rivalry, providing insight into population dynamics that may shape how visual signals are relayed to subsequent stages of visual processing.

neuroscience↗

Communication subspaces align with training in ANNs

Communication subspaces have recently been identified as a promising mechanism for selectively routing information between brain areas. In this study, we explored whether communication sub-spaces develop with training in artificial neural networks (ANNs) and explored differences across connection types. Specifically, we analyzed the subspace angles between activations and weights in ResNet-50 before and after training. We found that activations were more aligned to the weight layers after training, although this effect decreased in deeper layers. We also analyzed the angles between pairs of weight layers. We found that for all branching, direct, and skip connections, weight layer pairs were more geometrically aligned in trained versus untrained models throughout the entire network. These findings indicate that such alignment is essential for the proper functioning of deep networks and highlights the potential to enhance training efficiency through pre-alignment. In biological data, our results motivate further exploration into whether learning induces similar subspace alignment.

neuroscience↗