Ketamine-induced deficits in auditory and visual context-dependent processing in healthy volunteers: implications for models of cognitive deficits in schizophrenia

Cognitive deficits, particularly an impaired ability to form and use transient memory traces, are a core feature of schizophrenia and limit functional outcome. One electrophysiological marker of auditory sensory ("echoic") memory is the mismatch negativity (MMN), an event-related potential (ERP) that reflects automatic detection of deviant sounds. Patients with schizophrenia show reduced MMN and specific errors on a modified A–X Continuous Performance Test (AX-CPT) that indicate difficulty in forming and using brief task-relevant representations. Previous animal and human work implicates N-methyl-D-aspartate receptor (NMDAR) hypofunction in such abnormalities, and NMDAR antagonists abolish MMN in nonhuman primates. Umbricht and colleagues set out to test whether NMDAR blockade with subanesthetic ketamine reduces MMN generation in healthy humans and whether ketamine produces AX-CPT performance deficits similar to those seen in schizophrenia. The study therefore probes a mechanistic link between NMDAR function and transient-memory-dependent processing at both a sensory (MMN) and a cognitive (AX-CPT) level, using a within-subject, placebo-controlled pharmacological challenge.

This was a single-blind, placebo-controlled, within-subject crossover study in which each of 21 recruited healthy volunteers underwent both ketamine and placebo infusions on separate days in randomised, counterbalanced order; one subject was withdrawn for nausea, leaving 20 subjects (14 men, 6 women) in the final analysis (mean age 24.6 ± 2.9 years). Screening excluded current or past Axis I disorders, substance dependence/abuse, a family history of Axis I disorder to second-degree relatives, and medical illness. Intelligence and handedness were assessed, and informed consent obtained. The study was approved by a local ethics committee and sessions were conducted under constant supervision. Ketamine dosing comprised an intravenous bolus of 0.24 mg/kg over 5 minutes, a 5-minute pause, then a maintenance infusion nominally 0.9 mg/kg per hour with a 10% dose reduction every 15 minutes to stabilise blood levels. The placebo session followed the same schedule with saline/glucose infusion. ERP recordings were taken at three phases per session: baseline preinfusion, 20 minutes after the bolus during infusion, and after a 30-minute recovery period postinfusion. Subjects were blind to order. During ERP acquisition auditory stimuli (100 ms 1000 Hz standards, 100 ms 1500 Hz pitch deviants, 250 ms 1000 Hz duration deviants) were presented with a fixed sequence and 300 ms stimulus onset asynchrony while participants simultaneously performed a visual AX-CPT to divert attention from the sounds. The AX-CPT presented letters in pairs; subjects responded when an A cue was followed by an X target (AX). Stimulus pairs were presented in blocks with short (0.8 s) and long (4 s) interstimulus intervals (ISIs) intermixed; sequence probabilities were 70% AX and 10% each for BX, AY and BY. Electroencephalography used 28 scalp electrodes (10/20 montage plus mastoids) referenced to the nose, band-pass filtered 0.1–50 Hz and digitised at 500 Hz. Epochs comprised a 100 ms prestimulus baseline and 500 ms poststimulus window; artifact rejection and blink correction were applied and averages computed. Primary ERP measures at Fz were N1 and P2 to standards and MMN amplitudes/latencies obtained by subtracting standard from deviant waveforms, with pre-specified latency windows for peak detection. Behavioural ratings used the Brief Psychiatric Rating Scale (BPRS) and selected orientation items. Statistical analyses used repeated-measures ANOVAs with infusion (ketamine vs placebo) and time (baseline, infusion, postinfusion) as within-subject factors; MMN type (pitch vs duration) was an additional factor for MMN analyses. AX-CPT performance (hit rates and false alarms by type) included ISI (short vs long) as a within-subject factor. Contrasts and paired t tests were used for post hoc comparisons. The significance threshold was α = .05. The investigators also inspected MMN topography across electrode pairs to assess hemispheric effects.

Sample and tolerability: Of 21 enrolled subjects, one was excluded for nausea during ketamine, leaving 20 participants (14 men, 6 women) with mean verbal and performance IQs of 114 ± 11 and 117 ± 9. Ketamine produced the expected transient psychotomimetic and anergia effects: BPRS total score increased from 19.1 ± 1.3 at baseline to 33.3 ± 8.2 during infusion (infusion × time interaction F2,18 = 31.32, P < .001). Psychosis and anergia factor scores showed significant elevations; orientation was largely preserved. Mismatch negativity (MMN): Robust MMN responses to both pitch and duration deviants were present at baseline with maxima at Fz. Ketamine administration significantly reduced MMN peak amplitudes for both deviant types. Mean MMN-to-pitch amplitude fell by 1.47 µV (27%) and MMN-to-duration by 0.9 µV (21%) from session baselines. A repeated-measures ANOVA showed a significant infusion × time interaction (F2,18 = 4.57, P = .03). Ketamine also produced a modest but significant increase in MMN peak latency, driven mainly by delayed MMN for duration deviants (infusion × time interaction F2,18 = 9.96, P < .001); paired tests confirmed longer duration-MMN latency during ketamine versus baseline (t19 = 2.41, P = .03) and versus placebo (t19 = 3.56, P = .002). Topographic analyses across C3/C4, T3/T4 and P3/P4 showed no hemisphere-specific infusion effects. N1 and P2: Baseline N1 and P2 amplitudes/latencies were within expected ranges. Ketamine was associated with a small but significant increase in N1 peak amplitude (infusion × time interaction F2,18 = 4.70, P = .02); N1 amplitude during ketamine was larger than during placebo (t19 = -3.69, P = .002). N1 latency was unchanged. P2 amplitude increased slightly across sessions (effect of time F2,18 = 16.01, P < .001) but showed no specific ketamine effect, and P2 latency was unaffected. AX-CPT performance: One subject stopped the AX-CPT during ketamine, leaving 19 for behavioural analysis. Ketamine reduced hit rates (correct detection of AX sequences) at both ISIs (infusion × time interaction F2,17 = 10.02, P = .001), with post hoc tests showing significantly lower hit rates during ketamine versus baseline for short ISI (t18 = -4.63, P < .001) and long ISI (t18 = -4.60, P < .001). False alarms increased overall under ketamine, most markedly for BX sequences: a three-way infusion × time × false-alarm type interaction was significant (F4,15 = 3.33, P = .04). Contrasts showed a greater increase of BX errors compared with AY (F1,18 = 5.54, P = .03) and BY errors (F1,18 = 12.14, P = .003) during the infusion phase under ketamine versus placebo and baseline. BX error increases returned toward baseline after recovery. Signal-detection analyses (a') confirmed a significant performance decline during ketamine at both ISIs (short ISI t18 = -6.24, P < .001; long ISI t18 = -6.18, P < .001). The impairment did not differ by ISI, suggesting ketamine affected formation or use of transient cues rather than accelerated decay of maintained information.

Umbricht and colleagues interpret their findings as evidence that NMDARs are critically involved in human MMN generation and in cognitive operations that rely on transient memory traces. The ketamine-induced reduction of MMN amplitudes for both pitch and duration deviants, without an overall suppression of basic sensory ERPs of comparable latency, argues for a specific disruption of the neuronal mechanisms that produce MMN rather than a nonspecific weakening of ERP generators. The preservation of the approximately 100 ms latency difference between pitch and duration MMN across phases served as an additional validation that the affected components were genuine MMN waves. The investigators link the ketamine-induced MMN reduction to prior animal data showing abolition of MMN after NMDAR antagonists in auditory cortex, and to clinical observations of reduced MMN in recent-onset and chronic schizophrenia. They therefore propose that deficient NMDAR-dependent neurotransmission could underlie MMN deficits in schizophrenia. The discussion acknowledges some differences between ketamine effects and findings in patients: the study observed generally larger MMN amplitudes and a duration-MMN latency increase under ketamine that was not seen in the clinical sample cited, which could reflect methodological differences (deviant probability, stimulus parameters) or real distinctions between pharmacologically induced and illness-related deficits. Regarding mechanisms, the authors consider a predominantly local effect of ketamine on primary auditory cortex but also note the potential contribution of impaired top-down control from prefrontal cortex. Ketamine is known to increase prefrontal metabolic activity and prefrontal lesions in humans can reduce MMN and increase N1—patterns partially mirrored here—so a combination of local cortical disruption and altered prefrontal modulation is proposed. On the AX-CPT, ketamine produced a selective increase in BX errors—responses reflecting failure to form or use a transient representation of the cue—matching the error profile commonly reported in schizophrenia. Because the impairment did not worsen at the long ISI, the authors argue the deficit implicates initial formation or utilisation processes rather than faster decay of maintained information. Comparisons with other pharmacological manipulations suggest the effects are unlikely to be explained by nonspecific attentional disruption or actions at non-NMDA receptors: agents such as methylphenidate or an ACTH fragment altered attention-dependent ERPs without affecting MMN, whereas other NMDA antagonists (nitrous oxide, ethanol) reduce MMN. Moreover, although ketamine increases glutamate release and serotonin 5-HT2A agonists can also elevate glutamate, a study with psilocybin (a 5-HT2A agonist) did not reduce MMN despite similar behavioural effects, supporting an NMDA-related mechanism. The authors acknowledge dose differences and paradigm variations as factors that may explain discordant findings in the literature. Limitations discussed include differences between experimental parameters and clinical populations that may account for some discrepancies, and the fact that ketamine is a racemic mixture with off-target affinities albeit much lower than for NMDARs; the investigators consider it unlikely that non-NMDA receptor actions account for the principal effects. They also note the study probes acute pharmacological modelling in healthy subjects and thus provides indirect rather than direct evidence about pathophysiology in schizophrenia. Implications raised by the authors are that NMDAR-related dysfunction may contribute to deficits in transient memory traces observed in schizophrenia across sensory and cognitive levels, supporting the glutamatergic/NMDA hypothesis of the disorder.

The study concluded that subanesthetic ketamine impairs MMN generation and produces specific AX-CPT deficits in healthy volunteers that closely resemble deficits observed in schizophrenia. These effects are consistent with impairments in formation and use of transient memory traces, and the findings support a role for deficient NMDAR-dependent neurotransmission in such deficits in schizophrenia, lending support to the glutamatergic/NMDA model of the disorder.