Semantic influence on visual working memory of object identity and location

Not all objects are created equal: The object benefit in visual working memory is supported by greater recollection-like memory, but only for memorable objects

Visual working memory is thought to have a fixed capacity limit. However, recent evidence suggests that a greater number of real-world objects than simple features (i.e., colors) can be maintained, an effect termed the object benefit. Here, we examined whether this object benefit in visual working memory is due to qualitatively different memory processes employed for meaningful stimuli compared to simple features. In online samples of young adults, real-world objects were better remembered than colors, had higher measures of recollection, and showed a greater proportion of high-confidence responses (Exp. 1). Objects were also remembered better than their scrambled counterparts (Exp. 2), suggesting that this benefit is related to semantic information, rather than visual complexity. Critically, the specific objects that were likely to be remembered with high confidence were highly correlated across experiments, consistent with the idea that some objects are more memorable than others. Visual working memory performance for the least-memorable objects was worse than that of colors and scrambled objects. These findings suggest that real-world objects give rise to recollective, or at least high-confidence, responses at retrieval that may depend on activation of semantic features, but that this effect is limited to certain objects.

Can templates-for-rejection suppress real-world affective objects in visual search?

Previous evidence has suggested that feature-based templates-for-rejection can be maintained in working memory to suppress matching features in the environment. Currently, this effect has only been demonstrated using abstract neutral shapes, meaning that it is unclear whether this generalizes to real-world images, including aversive stimuli. In the current investigation, participants searched amongst an array of real-world objects for a target, after being precued with either a distractor template, target template, or a no template baseline. In Experiment 1, where both distractor and target template cues were presented randomly on a trial-by-trial basis, there was moderate evidence of increased capture by aversive distractors after the distractor template cue. In Experiment 2a, however, when distractor templates were the only available cue and more time was given to encode the cue features, there was moderate evidence of effective distractor inhibition for real-world aversive and neutral stimuli. In Experiment 2b, when the task required a slower more effortful comparison of target features to stereotypical object representations, there was weaker evidence of inhibition, though there was still modest evidence suggesting effective inhibition of aversive distractors. A Bayesian meta-analysis revealed that across Experiment 2, aversive distractors showed strong cumulative evidence of effective inhibition, but inconsistent inhibition for neutral distractors. The results are interpreted from a rational search behaviour framework, which suggests that individuals utilize informative cues when they enable the most beneficial strategy and are accessible, and apply these to distractors when they cause sufficient disruption, either to search speed or emotional state.

Representation and computation in visual working memory

The ability to sustain internal representations of the sensory environment beyond immediate perception is a fundamental requirement of cognitive processing. In recent years, debates regarding the capacity and fidelity of the working memory (WM) system have advanced our understanding of the nature of these representations. In particular, there is growing recognition that WM representations are not merely imperfect copies of a perceived object or event. New experimental tools have revealed that observers possess richer information about the uncertainty in their memories and take advantage of environmental regularities to use limited memory resources optimally. Meanwhile, computational models of visuospatial WM formulated at different levels of implementation have converged on common principles relating capacity to variability and uncertainty. Here we review recent research on human WM from a computational perspective, including the neural mechanisms that support it.

Learning-Induced Plasticity Enhances the Capacity of Visual Working Memory

Visual working memory (VWM) is limited in capacity, though memorizing meaningful objects may refine this limitation. However, meaningful and meaningless stimuli typically differ perceptually, and objects' associations with meaning are usually already established outside the laboratory, potentially confounding experimental findings. Here, in two experiments with young adults (N = 45 and N = 20), we controlled for these influences by having observers actively learn associations of (for them) initially meaningless stimuli: Chinese characters, half of which were consistently paired with pictures of animals or everyday objects in a learning phase. This phase was preceded and followed by a (pre- and postlearning) change-detection task to assess VWM performance. The results revealed that short-term retention was enhanced after learning, particularly for meaning-associated characters, although participants did not quite reach the accuracy level attained by native Chinese observers (young adults, N = 20). These results thus provide direct experimental evidence that participants' VWM of objects is boosted by them having acquired a long-term-memory association with meaning.

A Model of Semantic Completion in Generative Episodic Memory

Many studies have suggested that episodic memory is a generative process, but most computational models adopt a storage view. In this article, we present a model of the generative aspects of episodic memory. It is based on the central hypothesis that the hippocampus stores and retrieves selected aspects of an episode as a memory trace, which is necessarily incomplete. At recall, the neocortex reasonably fills in the missing parts based on general semantic information in a process we call semantic completion. The model combines two neural network architectures known from machine learning, the vector-quantized variational autoencoder (VQ-VAE) and the pixel convolutional neural network (PixelCNN). As episodes, we use images of digits and fashion items (MNIST) augmented by different backgrounds representing context. The model is able to complete missing parts of a memory trace in a semantically plausible way up to the point where it can generate plausible images from scratch, and it generalizes well to images not trained on. Compression as well as semantic completion contribute to a strong reduction in memory requirements and robustness to noise. Finally, we also model an episodic memory experiment and can reproduce that semantically congruent contexts are always recalled better than incongruent ones, high attention levels improve memory accuracy in both cases, and contexts that are not remembered correctly are more often remembered semantically congruently than completely wrong. This model contributes to a deeper understanding of the interplay between episodic memory and semantic information in the generative process of recalling the past.

Conceptual knowledge shapes visual working memory for complex visual information

Human visual working memory (VWM) is a memory store people use to maintain the visual features of objects and scenes. Although it is obvious that bottom-up information influences VWM, the extent to which top-down conceptual information influences VWM is largely unknown. We report an experiment in which groups of participants were trained in one of two different categories of geologic faults (left/right lateral, or normal/reverse faults), or received no category training. Following training, participants performed a visual change detection task in which category knowledge was irrelevant to the task. Participants were more likely to detect a change in geologic scenes when the changes crossed a trained categorical distinction (e.g., the left/right lateral fault boundary), compared to within-category changes. In addition, participants trained to distinguish left/right lateral faults were more likely to detect changes when the scenes were mirror images along the left/right dimension. Similarly, participants trained to distinguish normal/reverse faults were more likely to detect changes when scenes were mirror images along the normal/reverse dimension. Our results provide direct empirical evidence that conceptual knowledge influences VWM performance for complex visual information. An implication of our results is that cognitive scientists may need to reconceptualize VWM so that it is closer to "conceptual short-term memory".