Ketamine rotates mnemonic trajectories rather than erasing them

Question: How do NMDA receptor hypofunction and ketamine alter delay-period persistent activity and neural state-space trajectories in prefrontal and parietal cortex during working memory tasks?

Answer

Across the retrieved papers, the evidence supports a reconfiguration account rather than a loss-of-activity account: NMDAR antagonism does not abolish delay-period activity but changes what that activity carries and how it moves through state space. In parietal cortex the single most direct observation is geometric — subanesthetic ketamine rotated delay-period activity in LIP away from its pre-drug configuration while flattening single-neuron anticipatory ramps [10]. In prefrontal cortex the convergent signature is more firing with less information: LPFC firing rates rose while saccade-related signals carrying corollary discharge were suppressed and eye-movement discriminability fell [2], and human fMRI showed broadened spatial tuning with attenuated task-evoked activation across frontal and parietal areas, with tuning breadth rather than activation amplitude tracking individual memory performance [1]. Mechanistically, the degradations reported are consistency-preserving but separability-reducing: diminished direction-selective population coding after adolescent NMDAR hypofunction [3], shorter intrinsic timescales and reduced co-activation similarity under reduced excitatory-to-excitatory coupling [13], and a control geometry with weak access to item-differentiating directions [6]. Because delay activity is intrinsically non-stationary even without drugs [16], trajectory- and subspace-level measures, not single-unit rates or delay-versus-baseline contrasts, are the only ones that can adjudicate between these accounts [8].

Ketamine rotates delay-period population activity instead of silencing it

In macaque LIP, graded subanesthetic ketamine flattened single-neuron anticipatory ramping but at the population level produced a rotation of delay-period activity in neural state space away from the pre-ketamine configuration, in parallel with a dose-dependent dilation of estimated time [10]. This establishes that the trajectory can remain present and informative in a transformed geometry while its temporal parameterization is distorted, so "reduced persistent activity" is the wrong description of the drug effect even where behavior degrades.

Prefrontal signatures are elevated firing with degraded selectivity, not hypoactivity

Ketamine raised overall firing rates in LPFC while selectively suppressing saccade-related responses that carry corollary discharge, reducing discriminability of eye-movement signals and impairing performance [2]. Human pharmacological fMRI similarly found attenuated task-evoked activation across visual, parietal, and frontal areas, but ketamine-induced broadening of neural spatial tuning was more consistent across individuals and regions than activation change and correlated with memory precision [1]. A meta-analysis of working-memory fMRI complicates any simple hypoactivity story: chronic psychosis showed higher, not lower, activation in medial prefrontal cortex and rostral anterior cingulate alongside lower cerebellar, precuneal, and thalamic activation, whereas early psychosis showed lower dorsal anterior cingulate, caudate, and inferior frontal activation [20].

The degradation is best indexed by separability along item-differentiating dimensions

A stabilized supralinear network model implementing NMDAR hypofunction on pyramidal neurons reproduced shorter intrinsic neural timescales and reduced co-activation pattern similarity through reduced excitatory-to-excitatory strength, with the timescale–similarity correlation confirmed in chronic-ketamine mice and in schizophrenia [13]. In a spiking frontoparietal model with an intraparietal ring attractor and dlPFC actuator, the control geometry was strongly anisotropic and dominated by a common mnemonic direction, with much weaker access to item-differentiating directions [6]. Mean-field analysis shows the persistent-gamma regime sits closer to a Hopf boundary than the stimulus-locked regime, making maintenance intrinsically more fragile to E→I, I→E, and E→I variability changes [19]. Separability loss can also arise without an interneuron-centric E/I shift: repeated adolescent ketamine weakened mediodorsal thalamus→dmPFC presynaptic release, diminished direction-selective population coding in dmPFC, and chemogenetic strengthening of that projection restored both selectivity and behavior [3].

Delay activity is already non-stationary, so trajectory metrics are mandatory

High-density recordings from lateral prefrontal and posterior parietal cortex showed persistent delay activity in population averages while individual prefrontal neurons had silent periods indistinguishable from chance, with weakly correlated temporal fluctuations in mnemonic information across the two regions indicating partially synchronized off-states [16]. Perturbation sensitivity itself is dimension-specific: dlPFC microstimulation altered neurons and time-evolving dimensions but left stable working-memory dimensions and behavior largely intact [15]. Rotational state-space dynamics that return spiking to pre-distraction patterns, fuller on correct trials and coupled to traveling waves, supply a ready metric for trajectory integrity [12]. Identification of memory neurons required contrasting delay periods across tasks with different memory demands rather than delay-versus-baseline contrast within one task [8].

Attractor-geometry language gives the right vocabulary, and state changes can spare encoding

Data-constrained low-rank recurrent networks fitted to multi-session spiking revealed a manifold of fixed points during retention with attractor-like dynamics occupying position-specific subspaces [17]. In mouse mPFC, engagement state selectively modulated attractor dynamics within a memory-maintenance subspace while stimulus-evoked responses remained intact, and gradual engagement changes were predicted by continuous modulation of attractor geometry [14]. Prefrontal populations are organized into stereotyped subpopulations, one stimulus-driven and one maintenance-driven, rather than random mixed selectivity [11]. These results make "translation or rotation of an attractor manifold" a testable alternative to "weakened persistent activity," and predict that encoding can survive while delayed readout geometry shifts.

Not everything ketamine does is NMDAR-mediated, and NMDAR blockade alone may spare cognition

A selective NMDAR antagonist (CGS 19755) reproduced ketamine-like gamma-burst electrophysiology yet animals still performed a difficult memory task, and the antagonist lacked ketamine's delta-band modulation [5]. In resting MEG, GABAergic manipulation prolonged cortical timescales and shifted network-state occupancy while NMDAR modulation produced no significant changes [4]. In hippocampal recordings and network modeling, NMDAR blockade reduced overall activity and disrupted activity hierarchy, and removing the nonlinear NMDAR component flattened hierarchy and generated hypersynchronized slow oscillation [18]. Ketamine also destabilized network topology in human organoids by disconnecting high-connectivity "backbone" units [9]. In patients, anti-NMDAR encephalitis and schizophrenia both showed absent between-trial memory reactivations with impaired memory codes but intact perceptual codes, consistent with reduced short-term plasticity under NMDAR dysfunction [7].

Conflicts and gaps

The rate direction is unresolved: LPFC spiking increased [2] while task-evoked BOLD fell across frontal and parietal cortex [1], and chronic psychosis shows frontal hyperactivation relative to controls [20] — different signals, baselines, and chronicities are being conflated. Whether NMDAR manipulation changes cortical timescales at all is directly contested [13] versus [4]. The locus of the deficit is disputed between pyramidal recurrent excitation [13], presynaptic thalamofrontal input [3], and interneuron-dependent E/I fragility [19]. Critically, no retrieved study measures state-space trajectories in prefrontal cortex under ketamine; the one trajectory study is in parietal LIP during interval timing, not a working-memory delay [10], and no study aligns prefrontal and parietal subspaces within the same NMDAR-manipulated animal or participant. Region dissociation therefore remains unaddressed beyond the observation that tuning broadening appeared in both frontal and parietal areas [1].

What would change the answer

Simultaneous high-density PFC and PPC recordings in macaques during a memory-guided delay task, with pharmacokinetically clamped subanesthetic ketamine and cross-validated subspace alignment to test whether the delay trajectory rotates, contracts, or loses item separability, and with a selective NMDAR antagonist arm to separate NMDAR-dependent from ketamine-specific geometry changes [5]. Fitting low-rank RNNs to pre- and post-drug sessions would distinguish a moved fixed-point manifold from altered dynamics [17]. Applying the mid-delay distraction rotation assay under ketamine would test whether recovery rotations, not baseline activity, are the vulnerable quantity [12]. Pathway-specific rescue of mediodorsal thalamus→dmPFC release combined with trajectory measurement would test whether input restoration realigns the manifold [3]. In humans, 7T tuning-model fits with explicit subspace decoding during the delay would connect tuning breadth to trajectory geometry within the same participants [1].

References

[1] Rahmati, M., Moujaes, F., Suljic, N. P., Ji, J. L., Berkovitch, L., Lee, K., Fonteneau, C., Schleifer, C. H., Adkinson, B. D., Savic, A., Santamauro, N., Tamayo, Z., Diehl, C., Kolobaric, A., Flynn, M., Camarro, T., Curtis, C. E., Repovs, G., Fineberg, S. K., Morgan, P. T., Preller, K. H., Krystal, J. H., Murray, J. D., Cho, Y. T., Anticevic, A.. "Ketamine Alters Tuning of Neural and Behavioral Spatial Working Memory Precision". bioRxiv. 2025-04-14T00:00:00.000Z. URL: https://doi.org/10.1101/2025.02.10.637233

[2] Corrigan, B. W., Roussy, M., Luna, R., Gulli, R. A., Schall, J. D., Sachs, A. J., Palaniyappan, L., Martinez-Trujillo, J. C.. "Subanesthetic Ketamine Disrupts Predictive Signaling in the Prefrontal Cortex". bioRxiv. 2025-11-28T00:00:00.000Z. URL: https://doi.org/10.1101/2025.11.24.690250

[3] Yu, J., Choi, I. S., Kim, G. H., Bahn, S., Bae, S., Lee, T., Choi, J. H., Kwon, J. S., Kim, M., Lee, K. J., Rah, J.-C.. "Thalamofrontal synaptic weakening underlies short-term memory deficits from adolescent NMDAR hypofunction". bioRxiv. 2025-11-13T00:00:00.000Z. URL: https://doi.org/10.1101/2025.11.11.687794

[4] Dias Maile, A. A., Kohl, O., Ort, E., Froböse, M. I., Kurtenbach, H., Butz, M., Schreivogel, E., Schnitzler, A., Florin, E., Jocham, G.. "Role of GABA and NMDA receptors in shaping cortical timescales and large-scale network dynamics". bioRxiv. 2026-06-02T00:00:00.000Z. URL: https://doi.org/10.64898/2026.05.11.723797

[5] Acland, B. T., Palanca, B. J. A., Bijsterbosch, J. D., Snyder, L. H.. "Gamma-burst cortical activity in awake behaving macaques". bioRxiv. 2025-02-06T00:00:00.000Z. URL: https://doi.org/10.1101/2023.09.26.559594

[6] Gholaminejad, T., Holroyd, C. B.. "A Control-Theoretic Account of Cognitive Effort in Working Memory". bioRxiv. 2026-10-13T00:00:00.000Z. URL: https://doi.org/10.64898/2026.10.11.758303

[7] Heike Stein, Joao Barbosa, Diego Soldevilla, Mireia Rosa-Justicia, Alba Morató, Adrià Galan-Gadea, Laia Prades, Amaia Muñoz-Lopetegui, Helena Ariño, Eugenia Martinez-Hernandez. "Neural signatures of reduced serial dependence in anti-NMDAR encephalitis and schizophrenia". psyarxiv. 2026-01-24T17:37:10.187Z. URL: https://osf.io/kz73s_v2

[8] Huang, Y., Brosch, M.. "Linking working memory maintenance and readout in monkey sensory and prefrontal cortex". bioRxiv. 2026-03-16T00:00:00.000Z. URL: https://doi.org/10.64898/2026.03.13.711560

[9] Nikitina, A. A., Bustamante, C., Gifford, R., Camargo, C. M., Mejia-Cupajita, B., Kosik, K. S.. "Neuronal Population Effects of Ketamine on Human Brain Organoids". bioRxiv. 2026-03-10T00:00:00.000Z. URL: https://doi.org/10.64898/2026.03.09.710454

[10] Waters, J. R., Madeira, B., Shadlen, M. N., Shushruth, S.. "Neural correlates of ketamine-induced dilation of temporal expectation". bioRxiv. 2026-10-19T00:00:00.000Z. URL: https://doi.org/10.64898/2026.10.12.758574

[11] Pollan Hauer, N., Pesaran, B., Wimmer, K.. "Functional populations in prefrontal cortex related to working memory encoding and maintenance". bioRxiv. 2025-03-29T00:00:00.000Z. URL: https://doi.org/10.1101/2025.03.27.645288

[12] Batabyal, T., Brincat, S. L., Donoghue, J. A., Lundqvist, M., Mahnke, M. K., Miller, E. K.. "State-space trajectories and traveling waves following distraction". bioRxiv. 2025-06-27T00:00:00.000Z. URL: https://doi.org/10.1101/2024.02.19.581020

[13] Wengler, K., Hamm, J. P., Shymkiv, Y., Miller, K. D., Yuste, R., Horga, G.. "Convergent multidimensional neural phenotypes in human patients and schizophrenia-relevant mouse models identify reduced recurrent excitation in cortical circuits". medRxiv. 2026-09-24T00:00:00.000Z. URL: https://doi.org/10.64898/2026.09.22.26363726

[14] Osako, Y., Buschman, T. J., Sur, M.. "Internal state dynamically gates task-specific attractor dynamics in prefrontal cortex". bioRxiv. 2026-05-22T00:00:00.000Z. URL: https://doi.org/10.64898/2026.05.20.726585

[15] Soldado-Magraner, J., Minai, Y., Yu, B. M., Smith, M. A.. "Robustness of working memory to prefrontal cortex microstimulation". bioRxiv. 2025-01-14T00:00:00.000Z. URL: https://doi.org/10.1101/2025.01.14.632986

[16] Mozumder, R., Wang, Z., Dang, W., Zhu, J., Hammond, B., Machado, A., Constantinidis, C.. "Asynchronous firing and off-states in working memory maintenance". bioRxiv. 2025-10-13T00:00:00.000Z. URL: https://doi.org/10.1101/2025.10.12.681904

[17] Pals, M., Macke, J. H.. "Attractor dynamics underlie sequence working memory in data-constrained recurrent neural networks". bioRxiv. 2026-10-01T00:00:00.000Z. URL: https://doi.org/10.64898/2026.09.30.755565

[18] Zhou, Y., Wang, J. L., Qiu, L., Torpey, J., Wixson, J. G., Lyon, M., Chen, X.. "NMDA Receptors Control Activity Hierarchy in Neural Network: Loss of Control in Hierarchy Leads to Learning Impairments, Dissociation, and Psychosis". bioRxiv. 2024-12-16T00:00:00.000Z. URL: https://doi.org/10.1101/2023.01.06.523038

[19] Chung, D. W., Ermentrout, G. B.. "Differential Vulnerability of Stimulus-Locked and Persistent Gamma Oscillations: Implications in Schizophrenia". bioRxiv. 2026-05-29T00:00:00.000Z. URL: https://doi.org/10.64898/2026.05.29.728634

[20] Avram, M.-M., Bayly-Bureau, L., Livingston, N. R., Fusar-Poli, P., Kempton, M. J., Radua, J., Mehta, M. A., Modinos, G.. "Neurofunctional correlates of Working Memory across Psychosis Stages: A Systematic Review and Meta-analysis". medRxiv. 2026-01-14T00:00:00.000Z. URL: https://doi.org/10.64898/2026.01.12.26343386

How this survey was made

This survey answers one question: “How do NMDA receptor hypofunction and ketamine alter delay-period persistent activity and neural state-space trajectories in prefrontal and parietal cortex during working memory tasks?”

We keep our own collection of recent preprints and refresh it every day from bioRxiv, medRxiv, arXiv, PsyArXiv, SocArXiv and ChemRxiv. For each survey, we search that collection for recent work and keep the strongest matches: papers that use the same terms as the question, and papers on the same topic in different words.

The text was written from the titles and abstracts of those papers; the full texts were not read. Statements that draw on a particular paper are marked with a number in brackets. The reference list just above is the 20 papers whose work this text draws on, chosen from the 132 papers those searches returned; the rest of the shortlist is not printed. It is not a complete review of everything published on the question.

Treat this as a starting point and a reading list, not a peer-reviewed review. The searches behind this survey ran on October 21, 2026 and covered work from the previous two years.