Spatiotemporal dynamics across visual cortical laminae support a predictive coding framework for interpreting mismatch responses

Abstract

Context modulates neocortical processing of sensory data. Unexpected visual stimuli elicit large responses in primary visual cortex (V1)—a phenomenon known as deviance detection (DD) at the neural level, or “mismatch negativity” (MMN) when measured with EEG. It remains unclear how visual DD/MMN signals emerge across cortical layers, in temporal relation to the onset of deviant stimuli, and with respect to brain oscillations. Here we employed a visual “oddball” sequence—a classic paradigm for studying aberrant DD/MMN in neuropsychiatric populations—and recorded local field potentials in V1 of awake mice with 16-channel multielectrode arrays. Multiunit activity and current source density profiles showed that although basic adaptation to redundant stimuli was present early (50 ms) in layer 4 responses, DD emerged later (150–230 ms) in supragranular layers (L2/3). This DD signal coincided with increased delta/theta (2–7 Hz) and high-gamma (70–80 Hz) oscillations in L2/3 and decreased beta oscillations (26–36 Hz) in L1. These results clarify the neocortical dynamics elicited during an oddball paradigm at a microcircuit level. They are consistent with a predictive coding framework, which posits that predictive suppression is present in cortical feed-back circuits, which synapse in L1, whereas “prediction errors” engage cortical feed-forward processing streams, which emanate from L2/3.

Top-down input modulates visual context processing through an interneuron-specific circuit

“…Our recent studies show that only a subset of pyramidal neurons (PYRs) in the V1 exhibit DD and that these neurons are enriched in layer 2/3 (L2/3) of the cortex, 4 , 6 consistent with theoretical models 7 and empirical reports of visuomotor prediction errors. 8 These DD signals were not present in early evoked potentials or firing responses in L4 of the V1, 2 , 4 , 6 suggesting that DD is not inherited from bottom-up/feedforward inputs. On the other hand, DD in the V1 depend on top-down input from a downstream cortical region, the anterior cingulate area (ACa), but axon terminals of ACa neurons in the V1 themselves did not exhibit DD.…”

“…Our results provide some evidence that oddball processing can be understood in this framework because top-down modulation of primary sensory regions is necessary for DD (putatively, prediction errors) and appears to signal the likely upcoming stimulus (putatively, predictions; Figures S4A - S4C ). Although our work here and elsewhere suggests that DD responses during the oddball paradigm fit many of the predictions of the predictive processing framework, including a strong presence in feedforward layers of the cortex, 4 , 6 a strong theta and gamma band signal, 6 increased bottom-up directional flow in cortical networks ( Figure 2G ), and hierarchical expression across distributed cortical networks, 33 whether such DD responses truly serve to “update” internal predictions is not strictly discernable from the current data. Thus, interpreting the current results as evidence of predictive processing is not yet fully warranted, and other theories may also hold weight.…”

“…29 For example, during active attentional tasks in non-human primates, top-down/feedback connections in the visual system exhibit activity in the alpha or beta domain (9–25 Hz), while bottom-up/feedforward and local connections exhibit oscillatory signatures in the delta/theta (3–8 Hz) and gamma (25–10 Hz) bands. 30 , 31 Across cortical regions, DD signals appear to occupy mainly delta/theta frequencies (3–8 Hz) in visual, 2 , 6 auditory, 32 and multimodal oddball paradigms, 33 supporting the hypothesis that DD signals are prediction errors that are “fed forward” in cortical networks. However, the frequency band the top-down modulation from ACa to V1 occupies in the oddball paradigm has not been established.…”

“…For one, ACa drive at 10 Hz modulated SSTs and PYRs in opposite directions ( Figure 5 ) despite both populations exhibiting SSA ( Figure 3 ). Second, based on our past work, SSA is likely inherited in the bottom-up inputs to L2/3 (and perhaps in thalamocortical inputs) because it is present in early evoked potentials in L4 and in PYR neurons in all layers 4 , 6 (unlike DD). Third, SSA was also not affected by suppression of local V1 interneurons like DD.…”

Prediction mismatch responses arise as corrections of a predictive spiking code

“…These experiments showed that also without stimulus adaptation neurons can show strong prediction mismatch responses (PMRs) [6,7,8,9,10,11,12]. While it is still a matter of ongoing research how exactly PMRs arise, they often seem to be modulated by top-down inputs to the cortical circuits [4,13,14,15]. Together, these results suggest that top-down mediated predictions play a central role in cortical processing, and that PMRs are an important characteristic of their effect on neural dynamics.…”