[Insert Figure 3 here]
A few inconsistent MMN findings are worth mentioning. First, some people ‘recover’ without repeated psychotic episodes, and they do not exhibit an abnormal MMN (Kim et al., 2020). This highlights a key gap - few longitudinal studies have been conducted to ascertain how biomarkers change over time or with vacillating symptoms. Second, a small study in the subclinical population reported that high scorers on the suspiciousness subscale had larger MMN amplitudes than low scorers (Broyd et al., 2016). Thus, the current literature presents competing perspectives: 1) that increased symptom load reliably predicts abnormal MMN (e.g., (Claridge & Beech, 1995), but 2) in the subclinical population more subtle patterns of preserved and impaired MMN emerge. Clearly, more research is needed to address these gaps in our understanding.

Oculomotor and Low-Level Visual Deficits

Despite the wealth of data cataloging impaired auditory SM in schizophrenia and SSD, there is relatively little in the visual SM domain. This is surprising given the prevalence of oculomotor deficits in schizophrenia, suggesting that the visual system is affected. Before describing the studies that have reported on visual SM, we note research pinpointing oculomotor deficits in SSD. We address this issue first because oculomotor deficits lead to atypical sampling of the environment and these contribute to abnormal visual sensation. Oculomotor responses are disrupted along SSD with atypical performance in saccadic tasks (e.g., double-step, adaptation, anti-saccade) and smooth pursuit (O’Driscoll & Callahan, 2008; Thakkar & Rolfs, 2019; Wolf et al., 2021). A proposed mechanism is that abnormal corollary discharge (e.g., (Feinberg, 1978) disrupts knowledge about eye position, resulting in inaccurate eye movements with more interruptions during visual tracking. Findings in nonhuman primates and lesion studies implicate impaired signaling in the mediodorsal thalamus, which serves as an intermediary between subcortical eye movement control areas (e.g., superior colliculus) and frontal eye fields (Thakkar & Rolfs, 2019). Inactivating this pathway creates a mismatch between where an organism believes their eyes are directed and where they are actually looking and thus alters visual perception (Cavanaugh et al., 2016). In those with schizophrenia the abnormal visual scan patterns of faces and social settings contribute to aberrant social interactions (Patel et al., 2020). Oculomotor deficits contribute to higher order cognitive deficits because those with schizophrenia sample and perceive the environment abnormally.
Approximately ~60% of individuals with schizophrenia have distortions in visual perception (Phillipson & Harris, 1985) and >33% experience visual hallucinations (Silverstein & Lai, 2021). The range of perceptual deficits is broad and includes worse performance in assessments of contrast sensitivity (Harper et al., 2020), detection of contour (Keane et al., 2014), color (Fernandes et al., 2019), biological motion (Okruszek & Pilecka, 2017) but see (Keane, Peng, et al., 2018), faces (McCleery et al., 2015), and stronger afterimages (Thakkar et al., 2021). The extent of afterimage deficits is clinically relevant as they predict illness severity (Keane, Cruz, et al., 2018).
Similarly, accounts for visual deficits in schizophrenia spectrum disorders identify abnormalities in the anatomy of the retina (reviewed in reviewed in (Bernardin et al., 2017; Silverstein et al., 2020), as well as atypical network-level connectivity, both hypo- and hyper- connectivity, including cortical - medial temporal lobe abnormalities (reviewed in: (Silverstein & Lai, 2021). Together, the consistent oculomotor and low-level visual deficits suggest that processing from as early as the retina is abnormal in schizophrenia.

Visual Sensory Memory Biomarkers of Symptom Severity

Given the range of oculomotor and low-level visual deficits identified in SSD, it is surprising how little work has been conducted in iconic memory. Several studies investigating iconic memory decay in those with schizophrenia have reported no differences in schizophrenia (Hahn et al., 2011; Knight et al., 1978). We know of no research evaluating iconic SM in schizotypy in the general population .
Similarly, it is noteworthy that in comparison to the high volume of research on the auditory MMN, there is little research assessing the visual MMN in schizophrenia and in SSD. However, atypical visual MMN are documented and confirm that abnormal responses are not isolated to the auditory system. For example, the visual MMN is significantly reduced in individuals with schizophrenia in an oddball task using letters as stimuli (Neuhaus et al., 2013), visual motion direction (Urban et al., 2008) or horizontal compared to vertical grating patterns as oddballs (Farkas et al., 2015), with a similarly large effect size in group differences. The same pattern of reduced MMN amplitude has been characterized in studies using emotional faces (Csukly et al., 2013; She et al., 2017; Yin et al., 2018), in which the MMN reduction correlates with behavioral performance on an emotion recognition task (Csukly et al., 2013). One recent finding reports that individuals with schizophrenia showed normal visual MMN elicited by fearful faces but diminished amplitude MMNs elicited by neutral faces (Vogel et al., 2018). A new study has now identified larger visual MMN to happy compared to sad and neutral faces and found that MMN amplitude to the happy faces correlated with schizotypy scores, and interestingly, with measures of autism characteristics, highlighting the difficulty with MMN and specificity of diagnosis (Ford et al., 2022).
Another ERP that is linked to sensory memory is the P1. Visual P1 responses were significantly smaller in schizophrenia in response to low (but not high) spatial frequency gratings that was not related to visual acuity (Farkas et al., 2015). Although the P1 amplitudes did not correlate with symptoms, they were related to cognitive performance, as well as education level, duration of illness, and measures of global functioning (Farkas et al., 2015). Visual MMN has also been elicited in response to changes in the direction of moving dots and is significantly reduced in schizophrenia (Urban et al., 2008). Visual MMN reduction was again associated with global functioning, general symptoms, and medication dosage (Urban et al., 2008), highlighting the parallels between visual and auditory MMN. Despite these similarities across the modalities, we found no findings describing the visual MMN in other SSD or high schizotypy participants.
In summary, despite notable visual and oculomotor involvement in SSD, and a wealth of data demonstrating auditory SM deficits, little attention has been paid to visual SM specifically and no evidence of atypical iconic memory in SSD. Expanding visual SM research in the SSD population is needed to complement the auditory SM work and to deepen our understanding of visual SM contributions to other aspects of cognition.