Adding statistical regularity results in a global slowdown in visual search

The visual statistical learning overcomes scene dissimilarity through an independent clustering process

Contextual cueing is a phenomenon of visual statistical learning observed in visual search tasks. Previous research has found that the degree of deviation of items from its centroid, known as variability, determines the extent of generalization for that repeated scene. Introducing variability increases dissimilarity between multiple occurrences of the same repeated layout significantly. However, current theories do not explain the mechanisms that help to overcome this dissimilarity during contextual cue learning. We propose that the cognitive system initially abstracts specific scenes into scene layouts through an automatic clustering unrelated to specific repeated scenes, and subsequently uses these abstracted scene layouts for contextual cue learning. Experiment 1 indicates that introducing greater variability in search scenes leads to a hindering in the contextual cue learning. Experiment 2 further establishes that conducting extensive visual searches involving spatial variability in entirely novel scenes facilitates subsequent contextual cue learning involving corresponding scene variability, confirming that learning clustering knowledge precedes the contextual cue learning and is independent of specific repeated scenes. Overall, this study demonstrates the existence of multiple levels of learning in visual statistical learning, where item-level learning can serve as material for layout-level learning, and the generalization reflects the constraining role of item-level knowledge on layout-level knowledge.

Rethinking statistical learning as a continuous dynamic stochastic process, from the motor systems perspective

The brain integrates streams of sensory input and builds accurate predictions, while arriving at stable percepts under disparate time scales. This stochastic process bears different unfolding dynamics for different people, yet statistical learning (SL) currently averages out, as noise, individual fluctuations in data streams registered from the brain as the person learns. We here adopt a new analytical approach that instead of averaging out fluctuations in continuous electroencephalographic (EEG)-based data streams, takes these gross data as the important signals. Our new approach reassesses how individuals dynamically learn predictive information in stable and unstable environments. We find neural correlates for two types of learners in a visuomotor task: narrow-variance learners, who retain explicit knowledge of the regularity embedded in the stimuli. They seem to use an error-correction strategy steadily present in both stable and unstable environments. This strategy can be captured by current optimization-based computational frameworks. In contrast, broad-variance learners emerge only in the unstable environment. Local analyses of the moment-by-moment fluctuations, naïve to the overall outcome, reveal an initial period of memoryless learning, well characterized by a continuous gamma process starting out exponentially distributed whereby all future events are equally probable, with high signal (mean) to noise (variance) ratio. The empirically derived continuous Gamma process smoothly converges to predictive Gaussian signatures comparable to those observed for the error-corrective mode that is captured by current optimization-driven computational models. We coin this initially seemingly purposeless stage exploratory. Globally, we examine a posteriori the fluctuations in distributions' shapes over the empirically estimated stochastic signatures. We then confirm that the exploratory mode of those learners, free of expectation, random and memoryless, but with high signal, precedes the acquisition of the error-correction mode boasting smooth transition from exponential to symmetric distributions' shapes. This early naïve phase of the learning process has been overlooked by current models driven by expected, predictive information and error-based learning. Our work demonstrates that (statistical) learning is a highly dynamic and stochastic process, unfolding at different time scales, and evolving distinct learning strategies on demand.

Contextual cueing in co-active visual search: Joint action allows acquisition of task-irrelevant context

Repeatedly presenting a target within a stable search array facilitates visual search, an effect termed contextual cueing. Previous solo-performance studies have shown that successful acquisition of contextual memories requires explicit allocation of attentional resources to the task-relevant repeated contexts. By contrast, repeated but task-irrelevant contexts could not be learned when presented together with repeated task-relevant contexts due to a blocking effect. Here we investigated if such blocking of context learning could be diminished in a social context, when the task-irrelevant context is task-relevant for a co-actor in a joint action search mode. We adopted the contextual cueing paradigm and extended this to the co-active search mode. Participants learned a context-cued subset of the search displays (color-defined) in the training phase, and their search performance was tested in the transfer phase, where previously irrelevant and relevant subsets were swapped. The experiments were conducted either in a solo search mode (Experiments 1 and 3) or in a co-active search mode (Experiment 2). Consistent with the classical contextual cueing studies, contextual cueing was observed in the training phase of all three experiments. Importantly, however, in the "swapped" test session, a significant contextual cueing effect was manifested only in the co-active search mode, not in the solo search mode. Our findings suggest that social context may widen the scope of attention, thus facilitating the acquisition of task-irrelevant contexts.

Your face scares me: Effects of Perceptual load and Social Anxiety on processing of threatening and neutral faces

Social Anxiety Disorder is among the most widely studied psychiatric conditions. However, the role of attentional and emotional processes in the maintenance of the condition is still not well-established. This study addressed the impact of individual differences in Social Anxiety, by examining the effects of perceptual load and stimulus valence when processing faces vs objects, here used as distractors, within a letter-search task. In addition to RT and accuracy on the letter search task, heart rate, and skin conductance during the task and participants’ self-report emotional evaluation were assessed to help interpret performance effects. Results suggest that distractor stimuli that are either threatening or faces impair performance of high SA participants. Results demonstrate a hypervigilance for threatening faces in SA but indicate that this happens primarily when cognitive resources are available, that is, under low perceptual load.

Oculomotor anticipation reveals a multitude of learning processes underlying the serial reaction time task

Sequence learning is the cognitive faculty enabling everyday skill acquisition. In the lab, it is typically measured in speed of response to sequential stimuli, whereby faster responses are taken to indicate improved anticipation. However, response speed is an indirect measure of anticipation, that can provide only limited information on underlying processes. As a result, little is known about what is learned during sequence learning, and how that unfolds over time. In this work, eye movements that occurred before targets appeared on screen in an ocular serial reaction time (O-SRT) task provided an online indication of where participants anticipated upcoming targets. When analyzed in the context of the stimuli preceding them, oculomotor anticipations revealed several simultaneous learning processes. These processes influenced each other, as learning the task grammar facilitated acquisition of the target sequence. However, they were dissociable, as the grammar was similarly learned whether a repeating sequence inhabited the task or not. Individual differences were found in how the different learning processes progressed, allowing for similar performance to be produced for different latent reasons. This study provides new insights into the processes subserving sequence learning, and a new method for high-resolution study of it.

Neural Evidence Suggests Both Interference and Facilitation from Embedding Regularity into Visual Search

In this work, we relied on electrophysiological methods to characterize the processing stages that are affected by the presence of regularity in a visual search task. EEG was recorded for 72 participants while they completed a visual search task. Depending on the group, the task contained a consistent-mapping condition, a random-mapping condition, or both consistent and random conditions intermixed (mixed group). Contrary to previous findings, the control groups allowed us to demonstrate that the contextual cueing effect that was observed in the mixed group resulted from interference, not facilitation, to the target selection, response selection, and response execution processes (N2-posterior-contralateral, stimulus-locked lateralized readiness potential [LRP], and response-locked LRP components). When the regularity was highly valid (consistent-only group), the presence of regularity drove performance beyond general practice effects, through facilitation in target selection and response selection (N2-posterior-contralateral and stimulus-locked LRP components). Overall, we identified two distinct effects created by the presence of regularity: a global effect of validity that dictates the degree to which all information is taken into account and a local effect of activating the information on every trial. We conclude that, when considering the influence of regularity on behavior, it is vital to assess how the overall reliability of the incoming information is affected.

What can half a million change detection trials tell us about visual working memory?

Visual working memory (VWM) represents the surrounding world in an active and accessible state, but its capacity is severely limited. To better understand VWM and its limits, we collected data from over 3,800 participants in the canonical change detection task. This unique population-level data-set sheds new light on classic debates regarding VWM capacity. First, the result supported a view of VWM as an active process, as manifested by the fact that capacity estimates were not stable across set-sizes, but rather lower for the larger set-size. Another support for this notion came from the tight connection capacity estimates had with a measure of attentional control. Together, the data suggested that individual differences in capacity do not reflect only differences in storage-size, but differences in the efficiency of using this storage. Second, we found a response bias such that subjects are more likely to respond that the probed item changed, and this criterion bias was further shifted as the set-size increased. These findings are naturally explained by a slot-like theory arguing that when load exceeds capacity, certain items remain completely outside of VWM (instead of all items being represented in lower resolution), therefore causing subjects to perceive them as different from VWM contents even when they are unchanged. Additionally, we found that the pattern of d' also confirmed the predictions of a slot-like view of VWM, such that some items are represented with high fixed resolution and others are not represented at all, although this finding is based on two measures with very different underlying assumptions. We also discuss how flexible-resource views can accommodate these results. Moreover, comparing performance between the first and last trials demonstrated no evidence for proactive interference as the driving factor of capacity limitations. We provide further details regarding the distribution of individual capacity, the relations between capacity and demographic variables, and the spatial prioritization of the items.