Visual awareness is hypothesized to be intimately related to visual working memory (WM), such that information present in WM is thought to have necessarily been represented consciously. Recent work has challenged this longstanding view by demonstrating that visual stimuli rated by observers as unseen can nevertheless be maintained over a delay period. These experiments have been criticized, however, on the basis that subjective awareness ratings may contain response bias (e.g., an observer may report no awareness when in fact they had partial awareness). We mitigated this issue by investigating WM for visual stimuli that were matched for perceptual discrimination capacity (d′), yet which varied in subjective confidence ratings (so-called relative blindsight). If the degree of initial subjective awareness of a stimulus facilitates later maintenance of that information, WM performance should improve for stimuli encoded with higher confidence. In contrast, we found that WM performance did not benefit from higher visual discrimination confidence. This relationship was observed regardless of WM load (1 or 3). Insofar as metacognitive ratings (e.g., confidence, visibility) reflect visual awareness, these results challenge a strong relationship between conscious perception and WM using a paradigm that controls for discrimination accuracy and is less subject to response bias (since confidence is manipulated within subjects). Methodologically, we replicate prior efforts to induce relative blindsight using similar stimulus displays, providing a general framework for isolating metacognitive awareness in order to examine the function of consciousness.

When subjects become unconscious, there is a characteristic change in the way the cerebral cortex responds to perturbations, as can be assessed using transcranial magnetic stimulation and electroencephalography (TMS–EEG). For instance, compared to wakefulness, during non-rapid eye movement (NREM) sleep TMS elicits a larger positive–negative wave, fewer phase-locked oscillations, and an overall simpler response. However, many physiological variables also change when subjects go from wake to sleep, anesthesia, or coma. To avoid these confounding factors, we focused on NREM sleep only and measured TMS-evoked EEG responses before awakening the subjects and asking them if they had been conscious (dreaming) or not. As shown here, when subjects reported no conscious experience upon awakening, TMS evoked a larger negative deflection and a shorter phase-locked response compared to when they reported a dream. Moreover, the amplitude of the negative deflection—a hallmark of neuronal bistability according to intracranial studies—was inversely correlated with the length of the dream report (i.e., total word count). These findings suggest that variations in the level of consciousness within the same physiological state are associated with changes in the underlying bistability in cortical circuits.

Stimulus location is not always informative during visual short-term memory (VSTM) for nonspatial features. Nevertheless, there is considerable evidence for the automatic encoding and retention of location information, regardless of its task relevance. To explore the functional and neural bases of the representation of spatial context in VSTM for nonspatial information, functional magnetic resonance imaging was performed while subjects performed delayed recall for the orientation of individual stimuli. Stimulus location varied across trials, and although this information was irrelevant for task performance, multivariate pattern analysis decoding of stimulus location sustained across trials, and also the decoding strength, predicted the precision of the recall of orientation. The influence of spatial context on the representation of orientation was operationalized by comparing the orientation reconstructions with multivariate inverted encoding models (IEM) trained in location context-dependent vs. -independent data. Although orientation reconstructions were robust for both location-dependent and location-independent IEMs, they were markedly stronger for the former. Furthermore, the functional relevance of location context was demonstrated by the fact that only the location-dependent neural representations of stimulus orientation predicted recall precision. NEW & NOTEWORTHY Neural representation strength of stimulus location predicts the precision of visual short-term memory (VSTM) recall of nonspatial stimulus, even when this information is task irrelevant. Neural representations of nonspatial stimuli that incorporate location context are stronger than those that do not, and only the former representations are strongly linked to behavior. The contributions to nonspatial VSTM performance of the representation of location context are at least partly distinct from those of the representation of stimulus content.

Although humans can hold multiple items in mind simultaneously, the contents of working memory (WM) can be selectively prioritized to guide future behavior. We explored whether the “same-object” benefits in visual processing may also be observed in visual WM. fMRI data were collected while participants performed a multistep serial retrocuing task in which they first viewed two 2-D objects (coherently moving colored dots). During retention, an initial relevance cue then indicated whether only the first or only the second object (“object-relevant”), or only the color of both objects or only their direction of motion would be relevant for the remainder of the trial (“feature-relevant”). On “object-relevant” trials, the ensuing priority cues selected either one of the features (“color” or “direction”) bound to the relevance-cued object, whereas on “feature-relevant” trials, the priority cues selected one of the two relevance-cued features. Using multivariate inverted encoding models, we found a same-object benefit on object-relevant trials in occipitotemporal regions: On feature-relevant trials, the first priority cue triggered a strengthening of the neural representation of the cued feature and a concomitant weakening to baseline of the uncued feature, whereas on object-relevant trials, the cued item remained active but did not increase in strength and the uncued item weakened but remained significantly elevated throughout the delay period. Although the stimulus-specific representation in frontoparietal regions was weak and uneven, these regions closely tracked the higher order information of which stimulus category was relevant for behavior throughout the trial, suggesting an important role in controlling the prioritization of information in visual WM.

26Successful retrieval of an item from visual working memory (VWM) often requires an 27 associated representation of the trial-unique context in which that item was 28 presented. We dissociated the effects on fMRI signal of memory load versus context 29 binding by comparing nonspatial VWM for one oriented bar vs. three bars 30 individuated by their location on the screen vs. three items drawn from different 31 categories (orientation, color, and luminance), for which location context was 32 superfluous. Delay-period fMRI signal in frontal and parietal cortex was sensitive to 33 stimulus homogeneity rather than to memory load per se. Behavioral performance 34 revealed a broad range in swap errors, an index of the efficacy of context binding, 35and subjects were classified as high swap error or low swap error. During the delay 36 period, the strength of the representation of stimulus location in parietal cortex 37 predicted individual differences in swap errors. During recall, activity in occipital 38 cortex revealed two dissociable neural correlates of context binding: high swap-error 39 subjects allocated less spatial attention to the location of the probed item and more 40 spatial attention the location of non-probed items; high swap-error subjects also 41 represented the orientation of the probed item more weakly and the orientation of 42 nonprobed items more strongly. Our results suggest context binding is a computation 43 that influences all stages of VWM processing. 44 45 Significance Statement 46Although we often think of the contents of visual working memory (VWM) as 47representations of the items that need to be remembered, each item's trial-unique 48 context is also critical for successful performance. For example, if one observes a 49 red, then a black, then a blue car passing through an intersection, vivid memory for 50 the colors, alone, wouldn't allow one to execute the instruction "Follow the first of the 51 three cars that just drove by." Although manipulating load is commonly assumed to 52 isolate storage functions, requiring memory for multiple items drawn from the same 53 category also increases demands on the context binding needed to individuate these 54 items. This experiment tracked the influence of context binding on VWM stimulus 55 processing. 56 57 58 Individual differences in human visual working memory (VWM) capacity result from 60 several factors, including the strategic deployment of attention (Fukuda, Woodman, 61