Introduction

Vasoactive intestinal peptide-expressing (VIP) GABAergic interneurons occupy a distinctive position in the cortical inhibitory circuit. They are a numerically small subclass — under one in ten cortical interneurons across most rodent areas — yet they are recruited reliably by locomotion, arousal, reward, punishment, and top-down attention, and they have been written into circuit diagrams as the principal source of state-dependent disinhibition onto pyramidal cells via somatostatin (SST)–expressing interneurons Pi et al., 2013Lee et al., 2013Pfeffer et al., 2013Fu et al., 2014Karnani et al., 2016Tremblay et al., 2016. Across the past decade, this disinhibitory reading of VIP function has been generalised from auditory and frontal cortex to primary visual cortex, somatosensory cortex, hippocampus, and prefrontal–cingulate territories, and it now serves as the default circuit motif in network and learning models of cortex Fu et al., 2014Lee et al., 2013Kuchibhotla et al., 2016Williams & Holtmaat, 2019Hertäg & Sprekeler, 2019Litwin-Kumar et al., 2016Molino et al., 2017. The same generalisation, however, has surfaced empirical and theoretical conflicts that the field has not resolved: in several preparations VIP recruitment does not produce the expected SST suppression, and in several models the same connectivity supports either gain enhancement or paradoxical pyramidal suppression depending on weakly constrained parameters Pakan et al., 2016Dipoppa et al., 2018Yavorska & Wehr, 2021Molino et al., 2017Sanzeni et al., 2020.

This review takes that tension as its starting point. Rather than re-affirming the disinhibitory motif as a uniform property of the VIP subclass, the review asks what the evidence — molecular, developmental, morphological, electrophysiological, synaptic, in-vivo, oscillatory, computational, and translational — actually constrains, and where the boundaries of safe inference lie. The thesis the review defends is that VIP-IN function is reliably state-modulated and reliably disinhibitory at the subclass average across cortex, but that its sign, magnitude, gain regime, target compartment, and connectional asymmetry vary systematically with cortical area, behavioural context, and intra-subclass cell type, and that this variation is not noise but the substrate of VIP function Pi et al., 2013Pfeffer et al., 2013Fu et al., 2014Pakan et al., 2016Dipoppa et al., 2018Garrett et al., 2020Bigelow et al., 2019Anastasiades et al., 2021Bastos et al., 2023. Reading the corpus this way reframes the central question from “what does VIP do?” to “under what area×state×modulator conditions does each of the operating modes catalogued below engage?”.

The body of the review is structured as eleven sections that move from the molecular identity of the subclass outward to its translational implications. Molecular Identity and Transcriptomic Taxonomy reviews the transcriptomic and marker-gene evidence anchoring VIP within the caudal-ganglionic-eminence (CGE)–derived, 5-HT3AR+ branch of cortical inhibition and surveys current single-cell atlases that resolve the subclass into multiple t-types Rudy et al., 2010Tasic et al., 2016Tasic et al., 2018Hodge et al., 2019Yao et al., 2023Lee et al., 2023Gouwens et al., 2020. Developmental Origins and Postnatal Maturation reconstructs CGE specification, transcription-factor cascades, tangential migration, and postnatal maturation of the lineage Fishell & Rudy, 2011Lodato et al., 2011Vogt et al., 2014Mayer et al., 2018Batista-Brito et al., 2017Vucurovic et al., 2010. Morphological Diversity catalogues the bipolar/bitufted/multipolar morphological palette and the laminar distribution that constrains target choice Cauli et al., 1997Cauli et al., 2014Prönneke et al., 2015Apicella & Marchionni, 2022Emmenegger et al., 2018Gouwens et al., 2019. Intrinsic Electrophysiology assembles passive properties, Petilla firing-pattern membership, and the irregular- versus continuous-adapting taxonomic conflict, including the Patch-seq integration that pegs intrinsic phenotype to t-type Tremblay et al., 2016Gouwens et al., 2020.

Synaptic Properties and Connectivity inventories afferent and efferent connectivity — long-range cortico-cortical and thalamic input, neuromodulatory drive, and the asymmetric VIP→SST > VIP→PV/Pyr output that is the structural basis for the disinhibitory reading — and flags the layer- and area-specific exceptions to that asymmetry Pi et al., 2013Pfeffer et al., 2013Lee et al., 2013Karnani et al., 2016Karnani et al., 2016Wall et al., 2016Schneider-Mizell et al., 2025. Local Circuit Motifs and the Disinhibition Framework assembles those parts into the circuit motif itself, asks how reproducible the connectional asymmetry is across area and layer, and audits the optogenetic and chemogenetic studies that have driven the motif into the textbook Pi et al., 2013Lee et al., 2013Fu et al., 2014Karnani et al., 2016Cichon et al., 2017Williams & Holtmaat, 2019Veit et al., 2023. In Vivo Function During Behavior reviews calcium-imaging and electrophysiological recordings of VIP cells during locomotion, reward, punishment, and stimulus encoding, and assembles the cases in which the disinhibitory mechanism fails or runs in reverse Pakan et al., 2016Dipoppa et al., 2018Yavorska & Wehr, 2021Bigelow et al., 2019. VIP Interneurons Across Brain Regions lays out the cross-area heterogeneity directly: visual, auditory, somatosensory, frontal, and hippocampal VIP populations differ in afferent profile, output specificity, and behavioural recruitment Pi et al., 2013Dipoppa et al., 2018Bastos et al., 2023Anastasiades et al., 2021Tyan et al., 2014Gulyás et al., 1996Kawaguchi & Kubota, 1996.

Oscillatory Dynamics and Temporal Coordination examines temporal coordination — gamma, theta, Up-states, infraslow rhythms, and the predictive-coding/attention frame that ties VIP activity to inter-areal coordination Veit et al., 2023Hertäg & Sprekeler, 2020Hertäg & Clopath, 2021Shipp, 2016Francavilla et al., 2018. Species Differences, Human Relevance, and Disease works out cross-species conservation and divergence, primate expansion of CGE-derived diversity, and VIP-IN involvement in autism, Rett, Dravet, schizophrenia, and Alzheimer-related pathology Hodge et al., 2019Boldog et al., 2018Chartrand et al., 2023Bakken et al., 2021Bakken et al., 2021Krienen et al., 2020Goff et al., 2023McFarlan et al., 2024. Computational Models of VIP Circuit Function audits the rate-, spiking-, and learning-model literature, exposes the gain-regime conflicts the architectures generate, and argues that the present bottleneck is parameter identifiability rather than missing biology Litwin-Kumar et al., 2016Molino et al., 2017Veit et al., 2023Bos et al., 2025Hartung et al., 2024Tahvili et al., 2025Tahvili et al., 2025Reimann et al., 2026Sanzeni et al., 2020Lee et al., 2025Sabri & Batista-Brito, 2024.

Five cross-cutting tensions surface across these sections, and the review treats them as a single connected web rather than as section-local issues. The first is the discrete-versus-continuous structure of the VIP subclass: single-cell atlases recover discrete t-types in supervised clustering, while morphology, intrinsic phenotype, and Patch-seq cross-modal integration reveal graded, dataset-dependent boundaries within the subclass — a tension that is most explicit in Molecular Identity and Transcriptomic Taxonomy, Morphological Diversity, and Intrinsic Electrophysiology and that propagates into how downstream sections attribute function to “VIP” cells Tasic et al., 2016Tasic et al., 2018Prönneke et al., 2015Apicella & Marchionni, 2022Gouwens et al., 2020Schneider-Mizell et al., 2025Emmenegger et al., 2018. The second is the gain-regime ambiguity: optogenetic VIP activation has been read as subtractive, divisive, or mixed gain depending on area, layer, and analysis convention, and the rate models that fit the data span the same range Kuchibhotla et al., 2016Hertäg & Sprekeler, 2019Molino et al., 2017Litwin-Kumar et al., 2016Veit et al., 2023Bos et al., 2025Tahvili et al., 2025Tahvili et al., 2025. The third is the net-disinhibition versus paradoxical-effect conflict: the same VIP→SST→Pyr connectivity, parameterised inside or outside the inhibition-stabilised regime, supports either pyramidal facilitation or pyramidal suppression on VIP activation, so that “disinhibition” is a property of operating point and not of wiring alone Molino et al., 2017Sanzeni et al., 2020Reimann et al., 2026Hertäg & Sprekeler, 2019Tahvili et al., 2025. The fourth is the identifiability of the VIP→SST weight: in the four-population rate equations, VIP→SST and VIP→PV cross-couplings trade off against SST→Pyr and PV→Pyr terms in a way that connectomic priors alone do not constrain, and several published models with incompatible VIP→SST weights fit the same in-vivo data Hertäg & Sprekeler, 2019Litwin-Kumar et al., 2016Molino et al., 2017Veit et al., 2023Reimann et al., 2026Sabri & Batista-Brito, 2024. The fifth is cross-species translation: VIP-IN identity is conserved at the subclass level between rodent and primate cortex, but t-type proportions, laminar distribution, and human-specific morphologies (including L1 VIP PCDH20 and MC4R types) diverge non-trivially, so that mouse circuit conclusions transfer to human cortex with quantitative qualifications and disease-relevant uncertainty Hodge et al., 2019Boldog et al., 2018Chartrand et al., 2023Bakken et al., 2021Bakken et al., 2021Krienen et al., 2020Lee et al., 2023.

The review proceeds inductively from molecular ground truth to behavioural and translational implication, but the reader can also read it diagonally along any of the five tensions above. Forward and backward cross-references between Molecular Identity and Transcriptomic Taxonomy through Computational Models of VIP Circuit Function are dense and intentional: a claim about VIP firing in In Vivo Function During Behavior is only as strong as the molecular and morphological identity assigned to the recorded cells in Molecular Identity and Transcriptomic Taxonomy and Morphological Diversity Tasic et al., 2018Gouwens et al., 2020Prönneke et al., 2015Schneider-Mizell et al., 2025, and a claim about disinhibitory gain in Local Circuit Motifs and the Disinhibition Framework is only as strong as the synaptic-weight and sign of the model in Computational Models of VIP Circuit Function Hertäg & Sprekeler, 2019Molino et al., 2017Veit et al., 2023Reimann et al., 2026. The synthesis the review aims to deliver, summarised in Conclusion, is therefore not a unified theory of VIP function but a calibrated map of where the disinhibitory reading holds, where it does not, and what experiments and analyses would discriminate the remaining cases Pi et al., 2013Pakan et al., 2016Dipoppa et al., 2018Molino et al., 2017Reimann et al., 2026.