Working Memory Deficits in Schizophrenia Spectrum Disorders

Working memory (WM) deficits are a canonical deficit in people with schizophrenia (for reviews, (Forbes et al., 2009; Lett et al., 2014; Reichenberg, 2010), those with high schizotypy (Siddi et al., 2017), and those at high-risk of psychosis (Millman et al., 2022). WM impairments are broad and include verbal (reviewed in (Seabury & Cannon, 2020), visuospatial, and executive functioning aspects of WM (Barch & Ceaser, 2012; Forbes et al., 2009). It is worth noting that poor WM performance in people with schizophrenia is associated with a variety of neural differences including reduced gray matter volume (Du et al., 2022; Kochunov et al., 2022), abnormal (hypo- and hyper-) connectivity patterns (Ding et al., 2019; Du et al., 2022; Fryer et al., 2015; Hashimoto et al., 2010; Schutte et al., 2021; Seabury & Cannon, 2020; Unschuld et al., 2014), hypoactive clusters (Seabury & Cannon, 2020), and abnormal oscillations (Reilly et al., 2018). At a lower level, one account of the WM deficit in schizophrenia suggests that PFCinhibition is disrupted, leading to impaired WM (Meiron et al., 2022). PFC inhibition is thought to be driven by abnormal interneuron function, which in turn reduces gamma oscillation amplitude (Reilly et al., 2018); see also (Toader et al., 2020). A challenge in specifying underlying neural mechanisms of the WM deficit is that there is variability across samples with different patterns emerging in distinguishable subgroups (Rodriguez et al., 2019). Advances in neuroimaging provide important future paths for identifying biomarkers of working memory that reflect schizophrenia symptoms, and for producing therapies to mitigate symptoms. In the paragraphs below, we specify the findings specific to auditory and visual WM as a parallel to the sections on auditory and visual SM above.

Auditory Working Memory

Auditory WM is impaired across SSD. Deficits are known in schizophrenia, their first-degree relatives (Seidman et al., 2012; Seidman, Pousada-Casal, et al., 2016), and in the high-risk population (Higuchi et al., 2013; Rutschmann et al., 1980), as is verbal WM (Seabury & Cannon, 2020). Importantly, this deficit is apparent even in early-course schizophrenia (Gonzalez-Blanch et al., 2006) and is associated with abnormal neural responses (Leicht et al., 2015; Papageorgiou et al., 2001). Importantly, auditory WM performance predicts the presence of auditory hallucinations (Gisselgard et al., 2014; Jenkins et al., 2018). Therefore, similar to auditory SM, WM deficits may be related to symptoms rather than onset of psychosis.
Potential WM biomarkers are associated with altered neural function and patterns of neural connectivity. The neural correlates of WM deficits include the PFC (for a review, see (Hashimoto et al., 2010), and relevant sensory cortices (e.g., (Javitt & Sweet, 2015; Menon et al., 2001; Stevens et al., 1998). However, when investigating the neural correlates associated with abnormal WM functioning in schizophrenia, there is inconsistency. Several studies reported PFChyperactivity , whereas others reported the opposite (for a review see (Leitman et al., 2005). Other brain areas implicated in WM retrieval include parietal regions (for a review, see (Funahashi, 2017; Hamilton et al., 2018; Olson & Berryhill, 2009). Parietal function is also abnormal in schizophrenia particularly in those who hallucinate (Hashimoto et al., 2010; Wible et al., 2009) and hypoactive temporal lobe activity (Wible et al., 2009) correlates with schizophrenia symptom severity (Menon et al., 2001). More recent analyses investigating altered connectivity extend these regions-of-interest findings by identifying network dysfunction. For example, increased dorsolateral and medial PFC connectivity in those with schizophrenia and their nonclinical first-degree relatives, compared to controls (Unschuld et al., 2014), and within left superior frontal gyrus (Ding et al., 2019). Several distinct patterns of hyper- and hypo- connectivity are associated with deficits in verbal WM in people with schizophrenia – including altered salience network (SN), default mode network (DMN), and frontoparietal network (FPN) interactions (Rodriguez et al., 2019). Together, there is a network-level deficit in schizophrenia that impacts auditory WM.
Some evidence suggests auditory WM deficits are associated with other auditory-related symptoms such as auditory hallucinations (Geng et al., 2020; Gisselgard et al., 2014; Jenkins et al., 2018), and auditory processing deficits (Moschopoulos et al., 2021). For example, deficits in MMN to pitch deviants were related to deficits in prosody processing (non-verbal communication of emotion) in individuals with schizophrenia (Leitman et al., 2005) and P50 suppression in the auditory ERP correlated with WM performance (Hamilton et al., 2018). Network connectivity also identifies altered DMN activity associated with auditory hallucinations (Geng et al., 2020). Finally, there is an intriguing observation that the visual word form area engages during auditory perception and WM in people with schizophrenia (Herman et al., 2020), suggests that broadly altered structure and function underlies auditory WM function.
Worse WM performance is implicated in schizotypy (Park & McTigue, 1997; Siddi et al., 2017), particularly in those who displayed more positive and negative symptoms (Schmidt-Hansen & Honey, 2009), and was associated with worse communication abilities (Kerns & Becker, 2008). It has also been suggested that poorer working memory performance may be a biomarker for increased risk of developing psychosis due to the large effect sizes shown in schizotypy (Siddi et al., 2017). However, there is debate as to the robustness of the WM deficit (Lenzenweger & Gold, 2000) and whether the effect is due to impaired attentional ability, with some evidence indicating dissociable deficits (Marsh et al., 2017). In summary, abnormalities in auditory WM in schizophrenia are seen in diminished form in individuals with high schizotypy. This supports the idea that traits associated with schizophrenia are on a spectrum. Notably, compared to the auditory SM research, there is remarkably little research conducted on biomarkers of auditory WM deficits across the SSD.

Visual Working Memory in Schizophrenia Spectrum Disorders

A broad literature investigates visual WM deficits in people with schizophrenia and related disorders. Deficits in object, spatial and verbal WM are well-documented in review articles (e.g., (Barch, 2005; Haenschel & Linden, 2011; Piskulic et al., 2007). Briefly, across measures of accuracy, reaction time, and confidence, individuals with schizophrenia broadly perform less accurately, more slowly, and less confidently than neurotypicals. Impaired WM extends to the at-risk and schizotypy populations. In the at-risk population, two recent meta-analyses point to WM performance as reflecting symptom severity. Overall cognition, attention, processing speed, and WM were found to be significantly worse in those who were high-risk compared to healthy controls using the MATRICS Consensus Cognitive Battery (MCCB; (Zheng et al., 2018). Including both visual and auditory measures of working memory and symptoms using the Schedules for Clinical Assessment in Neuropsychiatry was able to distinguish between individuals diagnosed with schizophrenia, their unaffected first-degree relatives, and non-clinical individuals (Kent et al., 2004). Importantly, neural accounts for the visual WM impairment note the contributing factor of abnormal SM and atypical prefrontal function (e.g., (Haenschel & Linden, 2011).
A few biomarkers are associated with visual WM in SSD. There is a lower amplitude P100 (P1) responses over electrode site Oz in individuals with schizophrenia (Haenschel et al., 2009), and in those with high schizotypy (Koychev et al., 2010) to a visual delayed discrimination task, highlighting the contribution of early sensory processing on later cognition. Others report that schizophrenia participants have abnormal theta power during n-back performance, and reduced P100 in response to TMS pulses (Hoy et al., 2021). Importantly, atypical gamma oscillations can be observed in first-episode psychosis (Missonnier et al., 2020). Additional effects likely arise from gamma-band oscillatory activity during WM tasks that fails to increase with increased WM load in those with schizophrenia (Basar-Eroglu et al., 2007; Cho et al., 2006). A second biomarker of encoding-related WM deficit is the N2pc, which is also reduced in those with schizophrenia (Mayer et al., 2020). Broadly atypical frontal lobe activations in at risk youth (van Gool et al., 2022) and connectivity patterns are known across schizophrenia spectrum disorders (Briend et al., 2020; Schmidt et al., 2014), and at risk samples (Schutte et al., 2021).