A generic deviance detection principle for cortical On/Off responses, omission response, and mismatch negativity

Abstract

Neural responses to sudden changes can be observed in many parts of the sensory pathways at different organizational levels. For example, deviants that violate regularity at various levels of abstraction can be observed as simple On/Off responses of individual neurons or as cumulative responses of neural populations. The cortical deviance-related responses supporting different functionalities (e.g., gap detection, chunking, etc.) seem unlikely to arise from different function-specific neural circuits, given the relatively uniform and self-similar wiring patterns across cortical areas and spatial scales. Additionally, reciprocal wiring patterns (with heterogeneous combinations of excitatory and inhibitory connections) in the cortex naturally speak in favor of a generic deviance detection principle. Based on this concept, we propose a network model consisting of reciprocally coupled neural masses as a blueprint of a universal change detector. Simulation examples reproduce properties of cortical deviance-related responses including the On/Off responses, the omitted-stimulus response (OSR), and the mismatch negativity (MMN). We propose that the emergence of change detectors relies on the involvement of disinhibition. An analysis of network connection settings further suggests a supportive effect of synaptic adaptation and a destructive effect of Nmethyl-d-aspartate receptor (NMDA-r) antagonists on change detection. We conclude that the nature of cortical reciprocal wiring gives rise to a whole range of local change detectors supporting the notion of a generic deviance detection principle. Several testable predictions are provided based on the network model. Notably, we predict that the NMDA-r antagonists would generally dampen the cortical Off response, the cortical OSR, and the MMN.

Neural Substrates and Models of Omission Responses and Predictive Processes

“…The principle of a deviance detection circuit motif, which is replicated across the neocortex, is consistent with the notion of conservation of cortical microstructure across species and modalities (Douglas and Martin, 2004;Wang et al, 2010), and with the notion of a fundamental involvement of interneurons in the processing of stimuli according to context (Kuchibhotla et al, 2017;Lee et al, 2017;David, 2018). Such a circuit motif could accommodate a range of empirically informed models of local deviance detection (Schneider et al, 2018;Chien et al, 2019;Ross and Hamm, 2020) that rely on feedback and feedforward modulation of pyramidal output, mediated by interneuron populations with different dynamics.…”

“…Omission responses and MMN are both responses to deviants, with prominent generators in the sensory cortex that depend on stimulus modality (Bendixen et al, 2012). Thus, considering also the relative uniformity of cortical architecture, it is plausible that omission responses and MMN rely on a similar combination of SSA and genuine deviance detection and possibly on the same local circuitry (Chien et al, 2019). Karamürsel and Bullock (1999) hypothesized ''slow'' and ''fast'' omission responses to be generated by possibly similar, but ultimately separate and non-overlapping, mechanisms, especially because of the differential modulatory effect of attention on the responses obtained with either stimulation protocol.…”

Cortical Microcircuit Mechanisms of Mismatch Negativity and Its Underlying Subcomponents

“…Our synthesis of this data and hypothesized model (Figure 3) suggests that DD in layer 2/3 of neocortex depends on the presence of adaptation in a subset of the local network and/or in afferents from the thalamus or layer 4, but this requires additional investigation. In both humans and animals, it appears DD is facilitated by and functionally related to the degree of SSA (Taaseh et al, 2011;Chien et al, 2019). Together these results highlight the complementary but distinct nature of SSA and DD.…”

“…In contrast, DD, but seemingly not SSA, depends on signaling from SST interneurons, at least in the primary visual cortex, as inhibiting them abolishes DD while sparing SSA in principal neurons of primary visual cortex (Hamm and Yuste, 2016). Additionally, NMDA receptors may support oddball-driven responses and/or DD as blockade of NMDA receptors reduces MMN (Ehrlichman et al, 2008;Chen et al, 2015;Harms, 2016;Chien et al, 2019) but proper controls still need to be employed to determine if this reduction is due to altered DD or SSA (Harms, 2016). Again, there is conflicting evidence regarding the extent to which the NDMAR function contributes to SSA, making it difficult to discern its relationship to DD (Farley et al, 2010;Chen et al, 2015).…”

Augmenting Flexibility: Mutual Inhibition Between Inhibitory Neurons Expands Functional Diversity

“…To further illustrate this argument, we investigated the dynamics of a family of recurrent circuits we dubbed CRIREL (Coupled Recurrent Inhibition and Recurrent Excitation Loop). This type of structure is used across literature regarding a myriad of phenomena, such as decision making and deviance detection [47, 48].…”