Email Icon Facebook icon Twitter Icon GooglePlus Icon Contact

User Top Menu


Submitted by Tobias Bast on

Place cells are neurons in the mammalian hippocampus that fire if the animal is in a specific place. The discovery of place cells in rats in the 1970s was awarded a Nobel Prize in 2014. Place cells have been the focus of intense study, because place cell dynamics may play a central role in mediating the important memory functions associated with the hippocampus, including spatial and episodic memory (O’Keefe, 2014).

Zaremba et al. used two-photon calcium imaging of hippocampal place cell firing in head-fixed mice performing a goal-location learning ("goal-oriented learning," GOL) task on a treadmill, to study the relationship between place cell dynamics and behavior in wild-type (WT) mice and in Df(16)A+/− mice, an animal model of human 22q11.2 deletion syndrome. This study has implications for our fundamental understanding of how place cell dynamics may mediate goal-directed learning. Moreover, 22q11.2 deletion is one of the strongest identified risk factors for schizophrenia (schizophrenia) in humans and associated with the development of schizophrenia in up to one third of all cases, although this mutation accounts for < 1% of all schizophrenia cases (Karayiorgou et al., 2010). Therefore, the changes in place cell dynamics and task performance in the Df(16)A+/− mice may have implications for our understanding of cognitive deficits, in particular memory deficits, in schizophrenia. Indeed, the authors propose that "impaired hippocampal ensemble dynamics may be a central component of cognitive memory dysfunctions emerging from the 22q11.2 DS [deletion syndrome] and schizophrenia in general." My comments will focus on two challenges in applying the findings of this study to neuro-cognitive deficits in schizophrenia.

First, it remains to be clarified how behavioral deficits on the GOL task relate to cognitive deficits in schizophrenia. The task was carefully chosen, so as to facilitate the combination of hippocampal two-photon imaging of place cell activity with behavioral testing of goal-location learning and memory. The authors suggest "that the memory deficit revealed by the GOL task is reminiscent of episodic memory deficits." However, episodic memory is the memory of unique events and assays of episodic-like memory in animal models are, therefore, based on one-trial learning (Morris, 2001). In contrast, the GOL task involves multi-trial learning across several days. In the Introduction of their paper, the authors point to "a central role of [the hippocampus] in the pathophysiology of cognitive memory deficits in schizophrenia." Their experiment using functional inhibition of the hippocampus by muscimol shows that aspects of the GOL task require the hippocampus and may, thus, be suitable to probe hippocampal dysfunction relevant to the memory deficits in schizophrenia. However, the part of the GOL task demonstrated to require the hippocampus (i.e., the initial learning of the goal location on the treadmill, as well as the expression of this memory) is actually not disrupted in Df(16)A+/− mice, and it remains to be demonstrated that those aspects of the GOL task that are impaired in Df(16)A+/− mice require hippocampal function.

The main impairment in Df(16)A+/− mice was evident following a change in the environmental context: a context change disrupted performance in the Df(16)A+/− mice, whereas it left WT mice largely unaffected (Fig. 1C, Condition II). This increased context sensitivity is not really in line with a deficit in episodic-like memory, i.e. the memory of specific events and their spatio-temporal context (Morris, 2001), which – if at all – would rather be reflected by decreased context sensitivity. It is also decreased, rather than increased, context sensitivity that has typically been associated with hippocampal dysfunction (e.g., Maren & Holt, 2000; Good et al., 2007).

An interesting possibility, raised by the authors in their Discussion, is that the increased context sensitivity reflects "a misattribution of salience to irrelevant cues." This may be relevant to the schizophrenia-related phenotype of Df(16)A+/− mice, given that impaired salience allocation/selective attention has been implicated in psychotic symptoms, and could be confirmed using dedicated assays of salience allocation/selective attention (e.g., latent inhibition or blocking paradigms) (Gray et al., 1991; Kapur, 2003; Fletch and Frith, 2009). In addition, the authors suggested that Df(16)A+/− mice were impaired in learning a new goal location on the treadmill when tested in the familiar context again. However, Df(16)A+/− mice actually learnt the new goal location at a similar rate to WT mice, it was just that they started from a lower baseline (Fig. 1C, Condition III), which may again reflect a disruption caused by the context change.

Second, it remains to be clarified how the abnormal hippocampal place cell dynamics relate to key neural biomarkers of hippocampal dysfunction in schizophrenia. More specifically, metabolic hippocampal overactivity at rest, alongside impaired hippocampal recruitment during tasks normally requiring hippocampal activation, has emerged as a key feature of schizophrenia pathophysiology; this may reflect a deficient hippocampal inhibitory GABA system, as suggested by abnormal post-mortem markers of GABA function (Heckers and Konradi, 2015; Bast et al., 2017). Interestingly, using electrophysiological recordings, the authors showed a "dysregulation of hippocampal excitability during periods of rest" in Df(16)A+/− mice, as reflected by increased rate and power of hippocampal sharp-wave ripples (SWR) activity. In addition, previous experiments suggest reduced activity of hippocampal inhibitory GABA interneurons in these mice (Drew et al., 2011). SWRs reflect synchronous burst firing of hippocampal neuron populations (Csicsvari et al., 2000) and reduced hippocampal GABA function increases hippocampal burst firing (McGarrity et al., 2017). Therefore, the increased hippocampal SWR activity in Df(16)A+/− mice points to increased hippocampal activity at rest due to impaired GABA function, reminiscent of hippocampal pathophysiology in schizophrenia (Heckers and Konradi, 2015; Bast et al., 2017).

As discussed by the authors, the abnormal place cell dynamics may be a consequence of the aberrant SWR activity, although this remains to be clarified. Another possibility that should be considered is that aspects of the abnormal place cell dynamics in Df(16)A+/− mice are a consequence (rather than a cause) of abnormal GOL task behaviour. For example, as discussed by Zaremba et al. and supported by their data from WT mice, goals and motivational factors can be reflected in place cell firing. Therefore, changed place cell dynamics in Df(16)A+/− mice during later testing stages of the GOL task (i.e., during Conditions II and III) may partly reflect that these mice were less focused on the goal locations or that they received less reward.

Overall, the neurophysiological studies by Zaremba et al. offer intriguing insights into the relation of hippocampal place cell dynamics and behavior, as well as into alterations of this relation in Df(16)A+/− mice, a genetic mouse model relevant to schizophrenia. However, the significance of these alterations for memory deficits in schizophrenia remains to be clarified. In this respect, a more detailed characterization of the behavioral phenotype of Df(16)A+/− mice, using translational assays of clinically relevant neuro-cognitive functions, would be useful.


Cognitive deficits caused by prefrontal cortical and hippocampal neural disinhibition.
Bast T, Pezze M, McGarrity S
Br J Pharmacol. 2017 Oct; 174(19):3211-3225.

Ensemble patterns of hippocampal CA3-CA1 neurons during sharp wave-associated population events.
Csicsvari J, Hirase H, Mamiya A, Buzsáki G
Neuron. 2000 Nov; 28(2):585-94.

Evidence for altered hippocampal function in a mouse model of the human 22q11.2 microdeletion.
Drew LJ, Stark KL, Fénelon K, Karayiorgou M, MacDermott AB, Gogos JA
Mol Cell Neurosci. 2011 Aug; 47(4):293-305.

Perceiving is believing: a Bayesian approach to explaining the positive symptoms of schizophrenia.
Fletcher PC, Frith CD
Nat Rev Neurosci. 2009 Jan; 10(1):48-58.

Context- but not familiarity-dependent forms of object recognition are impaired following excitotoxic hippocampal lesions in rats.
Good MA, Barnes P, Staal V, McGregor A, Honey RC
Behav Neurosci. 2007 Feb; 121(1):218-23.

The neuropsychology of schizophrenia.
Gray JA, Feldon J, Rawlins JNP, Hemsley DR, Smith AD.
Behavioral and Brain Sciences. 14(1):1-84.

GABAergic mechanisms of hippocampal hyperactivity in schizophrenia.
Heckers S, Konradi C
Schizophr Res. 2015 Sep; 167(1-3):4-11.

Psychosis as a state of aberrant salience: a framework linking biology, phenomenology, and pharmacology in schizophrenia.
Kapur S
Am J Psychiatry. 2003 Jan; 160(1):13-23.

22q11.2 microdeletions: linking DNA structural variation to brain dysfunction and schizophrenia.
Karayiorgou M, Simon TJ, Gogos JA
Nat Rev Neurosci. 2010 Jun; 11(6):402-16.

The hippocampus and contextual memory retrieval in Pavlovian conditioning.
Maren S, Holt W
Behav Brain Res. 2000 Jun 01; 110(1-2):97-108.

Hippocampal Neural Disinhibition Causes Attentional and Memory Deficits.
McGarrity S, Mason R, Fone KC, Pezze M, Bast T
Cereb Cortex. 2016 Aug 22.

Episodic-like memory in animals: psychological criteria, neural mechanisms and the value of episodic-like tasks to investigate animal models of neurodegenerative disease.
Morris RG
Philos Trans R Soc Lond B Biol Sci. 2001 Sep 29; 356(1413):1453-65.

O´Keefe, J. (2014) Nobel Lecture: Spatial Cells in the Hippocampal Formation. Nobel Media AB. Web. 12 Sep 2017.

Submitted by Kenneth Hugdahl on

An often replicated finding in psychology is that what we see and hear of the world around us is colored by what we expect to see and hear, such that the actual perception of an object or a sound is the combination of the sensory input and our stored memories and prior expectations. Another way of expressing this is to say that our perceptions are the sum of bottom-up sensory input and top-down cognitive modulations. In perceptual and cognitive psychology, the expectations that people carry with them and which shape their perceptions are called “priors,” which can be more or less salient. Perceptual errors, like visual illusions, may then be seen as a mismatch between the sensory input and these “priors,” for example, the illusion that the fishing rod is broken if it is dipped into the water.

In their study of hallucinations, Powers et al. found that strong “priors” can induce beliefs about perceptual experience in the absence of a corresponding sensory input. In other words, strong priors can induce a conviction that a perceptual experience has an external cause, although there is none, as when individuals are convinced that they “hear a voice” although there is no one speaking to them. This is an auditory hallucination, and one of the most debilitating symptoms in schizophrenia. The advent of modern brain imaging techniques, and especially functional magnetic resonance imaging (fMRI), made it possible to record blood flow in the brain when subjects solved various cognitive tasks, or when they were hallucinating (see Kompus et al., 2011; Hugdahl, 2015; Sommer et al., 2008). A typical brain activation pattern in the temporal lobe language area during auditory hallucinations is seen in Figure 1.

Auditory Cortex 250

Although previous fMRI studies have shown how the brain is involved in hallucinations, it has not been known how the brain constructs such non-real percepts, and if they are caused by misinterpreted inner speech or by a perceptual mismatch between bottom-up perceptual and top-down cognitive influences.

The study by Powers et al. is, in this respect, an important contribution to a better understanding of the underlying mechanisms, with implications for more individualized treatments. The authors recruited four groups: patients with a diagnosis of psychosis who experienced “hearing voices”; patients who did not have such experiences; individuals without a diagnosis who “heard voices”; and individuals without a diagnosis who did not hear voices. In order to reveal the effect of priors on hallucinatory experiences, Powers et al. used an ingenious experimental paradigm based on Pavlovian, or classical, conditioning where they paired weak tones at sensory threshold concurrent on the presentation of a visual stimulus while they recorded blood flow changes in the brain with fMRI. Pairing the tones and the visual stimuli would then establish a conditioned, or learned, association between the two stimulus events. They then sometimes presented only the visual stimulus, and sometimes together with tones that were so weak that they could not be heard, and tested whether the subjects nevertheless believed that they had perceived a tone. If the subjects responded that they heard a tone and described how convinced they were, this constituted the foundation for the experience of an auditory hallucination (in the absence of a real perceptual stimulus).

The results showed that subjects in the groups that had previous experiences of auditory hallucinations, independent of a diagnosis, more often reported that they heard a tone, and were more convinced than the other two groups of their experience, and their brain responses also showed activation in networks that have previously been found in hallucinating individuals (see Figure 2 of Powers et al.). Thus, the results of the study by Powers et al. show how prior beliefs and expectations of a perceptual phenomenon, in this case an audio-visual association, can cause the experience of actually “hearing a voice” that does not exist. In this respect, the results also shed light on the ongoing discussion of theoretical models for auditory hallucinations, by providing strong arguments for the perceptual nature of the very strange phenomenon of being convinced of experiencing something that is simply not there.

Submitted by Mark Daly on

I think both play a role in each phenotype, and we need to invest more in analyses that span all types of variants. As they both have roughly 80 percent heritability, there is clearly a lot to learn from GWAS in autism, but to date the sample sizes have been much smaller than in schizophrenia. Conversely, with autism cases usually being children and collected with parents, the trio study design that enables de novo mutation discovery has been more deeply explored in autism but clearly has some signal also in schizophrenia (though seemingly a bit less, perhaps because most schizophrenia collections may not include children who were earlier diagnosed with a neurodevelopmental disorder such as developmental/intellectual disability or autism). As GWAS grows in autism and trio sequencing in schizophrenia, I would expect more discoveries and a better opportunity to draw insights from considering common and rare variations together.

Submitted by Chao Chen on

For different psychiatric diseases, should we focus more on de novo mutations/rare variants in autism and more on common variants in schizophrenia?

Submitted by Francis McMahon on

I was an early enthusiast for splitting based on clinical subtypes, but I think this has not been a very successful approach. When sample sizes are large enough―larger than we probably have so far for most disorders―some clinical subtypes may emerge, but I suspect that clinical symptoms are just too far removed from the genes to have a benefit that outweighs the loss of power from reduced sample size.

Submitted by Ellen Ovenden on

How important is clinical homogeneity in GWAS? Schizophrenia has many different subtypes/symptom domains that may be controlled by different pathways. Would grouping patients by these factors alter GWAS results?

Submitted by Adam Culbreth on

Disparate Neural Mechanisms of Effort-Based Reinforcement-Learning in Depression and Schizophrenia

This comment was co-written by Erin Moran.

Motivational deficits have long been associated with schizophrenia and major depressive disorder (MDD). These symptoms limit functioning of people with schizophrenia or MDD. However, current treatments for motivational impairment in schizophrenia and MDD are not effective for all patients. This limited effectiveness may stem from poor understanding of the mechanisms that give rise to these symptoms. Thus, work examining such mechanisms to elucidate similar/disparate areas of impairment in a transdiagnostic sample is of high relevance to the field.

In this recent article in the Journal of Neuroscience, Park and colleagues examined effort-based decision-making as one potential psychological mechanism for understanding motivational impairments in schizophrenia and MDD. Effort-based decision-making has become an extensively researched construct in the basic sciences with animal (Salamone et al., 2016) and human studies (Westbrook and Braver, 2016) generating a comprehensive picture of its associated neural processes, including dopamine systems, the ventral striatum, and the anterior cingulate cortex. This literature has provided models ripe for translation into clinical domains. Indeed, translational studies have found that individuals with MDD (Treadway et al., 2012; Yang et al., 2014) and schizophrenia (Gold et al., 2015) are less likely to exert effort to obtain monetary rewards compared to controls, and that this deficit correlates with motivational impairment across disorders. Further, studies have suggested roles of the striatum (Wolf et al., 2014; Huang et al., 2016; Yang et al., 2016), the cingulate cortex (Huang et al., 2016), and the dorsolateral prefrontal cortex (Wolf et al., 2014) in aberrant effort-based decision-making in patients with schizophrenia and MDD, suggesting potential preliminary neural mechanisms. However, although both MDD and schizophrenia seem to be associated with aberrant effort-based decision-making, it remains untested whether this seemingly similar deficit arises from similar or disparate neural mechanisms.

To this end, Park and colleagues collected fMRI data from individuals with schizophrenia (N = 19), MDD (N = 19), and healthy controls (N = 30) as they performed an effort-based reinforcement-learning task. On this task, participants were presented with a cue, indicating trial type. Following cue presentation, participants were presented with a start signal to press a button in reaction to this cue. Next, a work phase commenced where two light bulbs were presented and participants needed to quickly press a corresponding button to extinguish the light bulbs as they became lit. Finally, participants received feedback (i.e., gain, no gain, loss, no loss). Trial types varied by effort (i.e., high or low effort, depending on the number of button presses required) and reward type (i.e., positive or negative reinforcement). Behaviorally, the authors examined two dependent variables: reaction time of the cue response (cue/anticipation reaction time) and reaction time of the subject’s final light bulb switch-off (work/effort reaction time). Imaging data were analyzed by modeling the three task phases (anticipation, work, and feedback), as well as reward type and effort level in whole-brain and specific regions of interest analyses. Participants completed a resting-state scan before and after completing the effort task. Following scanning, participants completed a cue preference task, choosing their preferred cue type (i.e., neutral, positive, and negative) by making two alternative forced choices between trial types.

Park and colleagues found that controls showed significantly faster cue reaction time for positive low-effort trials and negative high-effort trials relative to neutral. In contrast, cue reaction time in MDD was only faster in the negative reward/high effort condition, and was not modulated by reward or effort level in schizophrenia. For work reaction time, Park and colleagues found individuals with MDD were slower on all four reinforcement conditions, but schizophrenia patients were slower on only the high-effort trials.

In regard to imaging, the researchers found that low-effort positive reinforcement trials and high-effort negative reinforcement trials were associated with greater putamen and medial orbital frontal cortex (OFC) activity in controls. In regard to group differences, they found that individuals with schizophrenia showed greater activity of the putamen during low-effort trials compared to control and MDD participants. Further, during the work phase, putamen activity was inversely correlated with motivational impairment during low-effort trials for those with schizophrenia, while in MDD putamen activation during the negative reward trials was inversely correlated with motivational impairment. With regard to resting-state functional connectivity, surprisingly, Park and colleagues found that subjects with schizophrenia showed largely similar functional connectivity compared to controls; however, they reported significantly reduced functional connectivity between the left putamen to right OFC in MDD relative to controls and schizophrenia patients. Moreover, the authors found that amotivation in schizophrenia was positively related to right nucleus accumbens and caudate nucleus (NAc/CHN) left medial OFC connectivity while amotivation in depression was positively correlated with left NAc/CHN-left medial OFC functional connectivity. This suggests a common relationship across disorders between amotivation and connectivity in the NAc/CHN-left medial OFC.

Park and colleagues are the first to take a transdiagnostic approach to examining the neural correlates of effort processing. However, interpretation of these results is limited by several factors. First, with regard to the behavioral data, it is not clear that the patients with schizophrenia learned the cues significantly above chance, and it does not appear that their preference for reward cues during the preference test was significantly greater than either the neutral or negative reinforcement cues. Learning of the cue is important to the interpretation of the results, which often involves contrasting various trial types. For example, the authors interpret the lack of modulation in cue reaction time by reward or effort level in individuals with schizophrenia as an overall lack of anticipatory capacity. However, an alternative explanation could be that patients did not sufficiently learn the cues and thus did not modulate reaction time to different cue presentations. This is an important point, given the mixed body of research pointing to the potential impairments of anticipatory response in schizophrenia (Frost and Strauss, 2016). Future work should rule out alternative explanations (e.g., impaired learning or understanding of task) when making claims about impaired anticipatory capacity in schizophrenia.

Second, it is unclear whether the authors’ measure of work RT assesses effort or a persistence/fatigue effect. Interestingly, individuals with MDD displayed slower work RT on all conditions, except neutral, while individuals with schizophrenia showed slowed work RT on high-effort trials compared to controls. It may be that the baseline motoric response of the patients is simply slower than the controls, particularly after repeated button presses, and it is likely that this slowness has no relation to effort processes. Further, this measure was not described in the original Croxson paper, limiting interpretation of the replicability of this measure.

With regard to the neuroimaging data, the authors often did not provide a comprehensive account of the statistical analyses, omitting effects for certain patient groups for various contrasts. Similarly, when interaction terms were discussed, most were not unpacked, leaving the reader unable to fully interpret the significance of the result or the direction of various effects. At times in the manuscript, it was unclear whether all diagnostic groups were included in a particular model or whether diagnostic groups were modeled separately. Such reporting makes the results section difficult to follow, and makes clear interpretation of the results challenging. Finally, the authors conducted a number of different tests with various contrasts, but found few significant group-level results. The small sample sizes of the patient groups limit the ability of the authors to make strong claims about negative findings, as these could simply reflect insufficient power to detect effects. As such, one useful topic for discussion would have been the replication or failure to replicate the initial Croxson and colleagues' study (Croxson et al., 2009). However, the authors did not discuss the original study outside of the methods section, and did not seem to replicate key findings of the Croxson paper, including effects in the ventral striatum and cingulate during effort/reward anticipation.

In summary, effort-based decision-making represents an intriguing contributory mechanism for motivational impairment across psychiatric disorders. Park and colleagues provide a novel study examining the neural correlates of aberrant effort-based decision-making in MDD, schizophrenia, and control participants. Such transdiagnostic samples are rare and allow for insights into seldom-posed questions including whether a similar behavioral deficit might arise from disparate neural mechanisms. However, interpretative challenges limit a full realization of the implications of their findings, which would have been helped by a more extensive reporting of statistical results, larger sample sizes (particularly in the patient groups), greater justification for the choice of behavioral measures and their construct validity as measures of effort-based decision-making, and a greater ability to rule out alternative explanations for the results. Taken together, these concerns raise questions about the authors’ interpretation of these data―an interpretation that could be misleading to the field. Future work is needed to provide replication and construct validity before we can conclude that there is clear evidence for differences between MDD and schizophrenia in the neural correlates of impaired effort-based decision-making.

Acknowledgements: We would like to thank Drs. Deanna Barch, James Waltz, and James Gold for helpful comments on an initial draft of this manuscript.

Submitted by Joshua Roffman on

One additional update: While change in negative symptoms did not correlate with change in mPFC thickness in the l-methylfolate study (n = 35), some new results from Walton, Ehrlich, and colleagues provide important additional context. In a paper published online in Psychological Medicine on May 26 (Walton et al., 2017), data from the ENIGMA consortium (n = 1,985) demonstrate a significant inverse relationship between negative symptom severity and cortical thickness in the very same medial prefrontal region where l-methylfolate treatment associated with cortical thickening. This additional evidence strengthens the case that LMF-induced changes in brain structure are meaningful for symptom improvement―a question that we will hopefully be able to address in larger, multisite studies. 

Submitted by Maureen Martin on

I wonder if response to methylfolate is associated with the presence of folate receptor autoantibodies in schizophrenia (Ramaekers et al., 2014), and if methylfolate response and folate receptor autoantibodies are markers of immune dysregulation similar to what is seen in PANDAS. If so, we could easily speculate on additional treatments that might be beneficial to these patients based on what is shown to help in PANDAS.

Submitted by Peter Uhlhaas on

This is a very intriguing paper that makes a number of important observations that are relevant for understanding the pathophysiology of schizophrenia. Firstly, diverse animal models converge on a dysfunction that implicates mainly neuronal ensembles rather than single-unit activity and thus highlights the importance of disturbances in large-scale neuronal dynamics as a critical component in schizophrenia. Secondly, these disturbances were brought about both by genetic risk factors as well as through blockade of NMDARs. Finally, this is one of the first preclinical studies that focuses on visual cortices that for a long time have been found to be fundamentally disturbed in schizophrenia. However, this aspect has been largely ignored in preclinical models of the disorder. Future research will need to verify in human neuroimaging data to what extent attractor dynamics are dysfunctional in schizophrenia (see commentary by Rolls) and how these observations can be linked to the important data presented in this paper.