The role of scene summary statistics in object recognition

Attention deployment in natural scenes: Higher-order scene statistics rather than semantics modulate the N2pc component

Which properties of a natural scene affect visual search? We consider the alternative hypotheses that low-level statistics, higher-level statistics, semantics, or layout affect search difficulty in natural scenes. Across three experiments (n = 20 each), we used four different backgrounds that preserve distinct scene properties: (a) natural scenes (all experiments); (b) 1/f noise (pink noise, which preserves only low-level statistics and was used in Experiments 1 and 2); (c) textures that preserve low-level and higher-level statistics but not semantics or layout (Experiments 2 and 3); and (d) inverted (upside-down) scenes that preserve statistics and semantics but not layout (Experiment 2). We included "split scenes" that contained different backgrounds left and right of the midline (Experiment 1, natural/noise; Experiment 3, natural/texture). Participants searched for a Gabor patch that occurred at one of six locations (all experiments). Reaction times were faster for targets on noise and slower on inverted images, compared to natural scenes and textures. The N2pc component of the event-related potential, a marker of attentional selection, had a shorter latency and a higher amplitude for targets in noise than for all other backgrounds. The background contralateral to the target had an effect similar to that on the target side: noise led to faster reactions and shorter N2pc latencies than natural scenes, although we observed no difference in N2pc amplitude. There were no interactions between the target side and the non-target side. Together, this shows that-at least when searching simple targets without own semantic content-natural scenes are more effective distractors than noise and that this results from higher-order statistics rather than from semantics or layout.

The Impact of Scene Context on Visual Object Recognition: Comparing Humans, Monkeys, and Computational Models

During natural vision, we rarely see objects in isolation but rather embedded in rich and complex contexts. Understanding how the brain recognizes objects in natural scenes by integrating contextual information remains a key challenge. To elucidate neural mechanisms compatible with human visual processing, we need an animal model that behaves similarly to humans, so that inferred neural mechanisms can provide hypotheses relevant to the human brain. Here we assessed whether rhesus macaques could model human context-driven object recognition by quantifying visual object identification abilities across variations in the amount, quality, and congruency of contextual cues. Behavioral metrics revealed strikingly similar context-dependent patterns between humans and monkeys. However, neural responses in the inferior temporal (IT) cortex of monkeys that were never explicitly trained to discriminate objects in context, as well as current artificial neural network models, could only partially explain this cross-species correspondence. The shared behavioral variance unexplained by context-naive neural data or computational models highlights fundamental knowledge gaps. Our findings demonstrate an intriguing alignment of human and monkey visual object processing that defies full explanation by either brain activity in a key visual region or state-of-the-art models.

Learned associations serve as target proxies during difficult but not easy visual search

The target template contains information in memory that is used to guide attention during visual search and is typically thought of as containing features of the actual target object. However, when targets are hard to find, it is advantageous to use other information in the visual environment that is predictive of the target's location to help guide attention. The purpose of these studies was to test if newly learned associations between face and scene category images lead observers to use scene information as a proxy for the face target. Our results showed that scene information was used as a proxy for the target to guide attention but only when the target face was difficult to discriminate from the distractor face; when the faces were easy to distinguish, attention was no longer guided by the scene unless the scene was presented earlier. The results suggest that attention is flexibly guided by both target features as well as features of objects that are predictive of the target location. The degree to which each contributes to guiding attention depends on the efficiency with which that information can be used to decode the location of the target in the current moment. The results contribute to the view that attentional guidance is highly flexible in its use of information to rapidly locate the target.

Context effects on object recognition in real-world environments: A study protocol

Background: The environments that we live in impact on our ability to recognise objects, with recognition being facilitated when objects appear in expected locations (congruent) compared to unexpected locations (incongruent). However, these findings are based on experiments where the object is isolated from its environment. Moreover, it is not clear which components of the recognition process are impacted by the environment. In this experiment, we seek to examine the impact real world environments have on object recognition. Specifically, we will use mobile electroencephalography (mEEG) and augmented reality (AR) to investigate how the visual and semantic processing aspects of object recognition are changed by the environment. Methods: We will use AR to place congruent and incongruent virtual objects around indoor and outdoor environments. During the experiment a total of 34 participants will walk around the environments and find these objects while we record their eye movements and neural signals. We will perform two primary analyses. First, we will analyse the event-related potential (ERP) data using paired samples t-tests in the N300/400 time windows in an attempt to replicate congruency effects on the N300/400. Second, we will use representational similarity analysis (RSA) and computational models of vision and semantics to determine how visual and semantic processes are changed by congruency. Conclusions: Based on previous literature, we hypothesise that scene-object congruence would facilitate object recognition. For ERPs, we predict a congruency effect in the N300/N400, and for RSA we predict that higher level visual and semantic information will be represented earlier for congruent scenes than incongruent scenes. By collecting mEEG data while participants are exploring a real-world environment, we will be able to determine the impact of a natural context on object recognition, and the different processing stages of object recognition.

Hierarchical organization of objects in scenes is reflected in mental representations of objects

The arrangement of objects in scenes follows certain rules ("Scene Grammar"), which we exploit to perceive and interact efficiently with our environment. We have proposed that Scene Grammar is hierarchically organized: scenes are divided into clusters of objects ("phrases", e.g., the sink phrase); within every phrase, one object ("anchor", e.g., the sink) holds strong predictions about identity and position of other objects ("local objects", e.g., a toothbrush). To investigate if this hierarchy is reflected in the mental representations of objects, we collected pairwise similarity judgments for everyday object pictures and for the corresponding words. Similarity judgments were stronger not only for object pairs appearing in the same scene, but also object pairs appearing within the same phrase of the same scene as opposed to appearing in different phrases of the same scene. Besides, object pairs with the same status in the scenes (i.e., being both anchors or both local objects) were judged as more similar than pairs of different status. Comparing effects between pictures and words, we found similar, significant impact of scene hierarchy on the organization of mental representation of objects, independent of stimulus modality. We conclude that the hierarchical structure of visual environment is incorporated into abstract, domain general mental representations of the world.

Auxiliary Scene-Context Information Provided by Anchor Objects Guides Attention and Locomotion in Natural Search Behavior

Successful adaptive behavior requires efficient attentional and locomotive systems. Previous research has thoroughly investigated how we achieve this efficiency during natural behavior by exploiting prior knowledge related to targets of our actions (e.g., attending to metallic targets when looking for a pot) and to the environmental context (e.g., looking for the pot in the kitchen). Less is known about whether and how individual nontarget components of the environment support natural behavior. In our immersive virtual reality task, 24 adult participants searched for objects in naturalistic scenes in which we manipulated the presence and arrangement of large, static objects that anchor predictions about targets (e.g., the sink provides a prediction for the location of the soap). Our results show that gaze and body movements in this naturalistic setting are strongly guided by these anchors. These findings demonstrate that objects auxiliary to the target are incorporated into the representations guiding attention and locomotion.

What to expect where and when: how statistical learning drives visual selection

While the visual environment contains massive amounts of information, we should not and cannot pay attention to all events. Instead, we need to direct attention to those events that have proven to be important in the past and suppress those that were distracting and irrelevant. Experiences molded through a learning process enable us to extract and adapt to the statistical regularities in the world. While previous studies have shown that visual statistical learning (VSL) is critical for representing higher order units of perception, here we review the role of VSL in attentional selection. Evidence suggests that through VSL, attentional priority settings are optimally adjusted to regularities in the environment, without intention and without conscious awareness.

Context effects on object recognition in real-world environments: A study protocol

Background: The environments that we live in impact on our ability to recognise objects, with recognition being facilitated when objects appear in expected locations (congruent) compared to unexpected locations (incongruent). However, these findings are based on experiments where the object is isolated from its environment. Moreover, it is not clear which components of the recognition process are impacted by the environment. In this experiment, we seek to examine the impact real world environments have on object recognition. Specifically, we will use mobile electroencephalography (mEEG) and augmented reality (AR) to investigate how the visual and semantic processing aspects of object recognition are changed by the environment. Methods: We will use AR to place congruent and incongruent virtual objects around indoor and outdoor environments. During the experiment a total of 34 participants will walk around the environments and find these objects while we record their eye movements and neural signals. We will perform two primary analyses. First, we will analyse the event-related potential (ERP) data using paired samples t-tests in the N300/400 time windows in an attempt to replicate congruency effects on the N300/400. Second, we will use representational similarity analysis (RSA) and computational models of vision and semantics to determine how visual and semantic processes are changed by congruency. Conclusions: Based on previous literature, we hypothesise that scene-object congruence would facilitate object recognition. For ERPs, we predict a congruency effect in the N300/N400, and for RSA we predict that higher level visual and semantic information will be represented earlier for congruent scenes than incongruent scenes. By collecting mEEG data while participants are exploring a real-world environment, we will be able to determine the impact of a natural context on object recognition, and the different processing stages of object recognition.

Is there such a thing as a 'good statistical learner'?

A growing body of research investigates individual differences in the learning of statistical structure, tying them to variability in cognitive (dis)abilities. This approach views statistical learning (SL) as a general individual ability that underlies performance across a range of cognitive domains. But is there a general SL capacity that can sort individuals from 'bad' to 'good' statistical learners? Explicating the suppositions underlying this approach, we suggest that current evidence supporting it is meager. We outline an alternative perspective that considers the variability of statistical environments within different cognitive domains. Once we focus on learning that is tuned to the statistics of real-world sensory inputs, an alternative view of SL computations emerges with a radically different outlook for SL research.

The role of contextual materials in object recognition

While scene context is known to facilitate object recognition, little is known about which contextual "ingredients" are at the heart of this phenomenon. Here, we address the question of whether the materials that frequently occur in scenes (e.g., tiles in a bathroom) associated with specific objects (e.g., a perfume) are relevant for the processing of that object. To this end, we presented photographs of consistent and inconsistent objects (e.g., perfume vs. pinecone) superimposed on scenes (e.g., a bathroom) and close-ups of materials (e.g., tiles). In Experiment 1, consistent objects on scenes were named more accurately than inconsistent ones, while there was only a marginal consistency effect for objects on materials. Also, we did not find any consistency effect for scrambled materials that served as color control condition. In Experiment 2, we recorded event-related potentials and found N300/N400 responses-markers of semantic violations-for objects on inconsistent relative to consistent scenes. Critically, objects on materials triggered N300/N400 responses of similar magnitudes. Our findings show that contextual materials indeed affect object processing-even in the absence of spatial scene structure and object content-suggesting that material is one of the contextual "ingredients" driving scene context effects.

Understanding Collections of Related Datasets Using Dependent MMD Coresets

Understanding how two datasets differ can help us determine whether one dataset under-represents certain sub-populations, and provides insights into how well models will generalize across datasets. Representative points selected by a maximum mean discrepancy (MMD) coreset can provide interpretable summaries of a single dataset, but are not easily compared across datasets. In this paper, we introduce dependent MMD coresets, a data summarization method for collections of datasets that facilitates comparison of distributions. We show that dependent MMD coresets are useful for understanding multiple related datasets and understanding model generalization between such datasets.

The meaning and structure of scenes

We live in a rich, three dimensional world with complex arrangements of meaningful objects. For decades, however, theories of visual attention and perception have been based on findings generated from lines and color patches. While these theories have been indispensable for our field, the time has come to move on from this rather impoverished view of the world and (at least try to) get closer to the real thing. After all, our visual environment consists of objects that we not only look at, but constantly interact with. Having incorporated the meaning and structure of scenes, i.e. its "grammar", then allows us to easily understand objects and scenes we have never encountered before. Studying this grammar provides us with the fascinating opportunity to gain new insights into the complex workings of attention, perception, and cognition. In this review, I will discuss how the meaning and the complex, yet predictive structure of real-world scenes influence attention allocation, search, and object identification.

Get Your Guidance Going: Investigating the Activation of Spatial Priors for Efficient Search in Virtual Reality

Repeated search studies are a hallmark in the investigation of the interplay between memory and attention. Due to a usually employed averaging, a substantial decrease in response times occurring between the first and second search through the same search environment is rarely discussed. This search initiation effect is often the most dramatic decrease in search times in a series of sequential searches. The nature of this initial lack of search efficiency has thus far remained unexplored. We tested the hypothesis that the activation of spatial priors leads to this search efficiency profile. Before searching repeatedly through scenes in VR, participants either (1) previewed the scene, (2) saw an interrupted preview, or (3) started searching immediately. The search initiation effect was present in the latter condition but in neither of the preview conditions. Eye movement metrics revealed that the locus of this effect lies in search guidance instead of search initiation or decision time, and was beyond effects of object learning or incidental memory. Our study suggests that upon visual processing of an environment, a process of activating spatial priors to enable orientation is initiated, which takes a toll on search time at first, but once activated it can be used to guide subsequent searches.

Contextual and Spatial Associations Between Objects Interactively Modulate Visual Processing

Much of what we know about object recognition arises from the study of isolated objects. In the real world, however, we commonly encounter groups of contextually associated objects (e.g., teacup and saucer), often in stereotypical spatial configurations (e.g., teacup above saucer). Here we used electroencephalography to test whether identity-based associations between objects (e.g., teacup-saucer vs. teacup-stapler) are encoded jointly with their typical relative positioning (e.g., teacup above saucer vs. below saucer). Observers viewed a 2.5-Hz image stream of contextually associated object pairs intermixed with nonassociated pairs as every fourth image. The differential response to nonassociated pairs (measurable at 0.625 Hz in 28/37 participants) served as an index of contextual integration, reflecting the association of object identities in each pair. Over right occipitotemporal sites, this signal was larger for typically positioned object streams, indicating that spatial configuration facilitated the extraction of the objects' contextual association. This high-level influence of spatial configuration on object identity integration arose ~ 320 ms post-stimulus onset, with lower-level perceptual grouping (shared with inverted displays) present at ~ 130 ms. These results demonstrate that contextual and spatial associations between objects interactively influence object processing. We interpret these findings as reflecting the high-level perceptual grouping of objects that frequently co-occur in highly stereotyped relative positions.

Automatic Attention Capture by Threatening, But Not by Semantically Incongruent Natural Scene Images

Visual objects are typically perceived as parts of an entire visual scene, and the scene’s context provides information crucial in the object recognition process. Fundamental insights into the mechanisms of context-object integration have come from research on semantically incongruent objects, which are defined as objects with a very low probability of occurring in a given context. However, the role of attention in processing of the context-object mismatch remains unclear, with some studies providing evidence in favor, but other against an automatic capture of attention by incongruent objects. Therefore, in the present study, 25 subjects completed a dot-probe task, in which pairs of scenes—congruent and incongruent or neutral and threatening—were presented as task-irrelevant distractors. Importantly, threatening scenes are known to robustly capture attention and thus were included in the present study to provide a context for interpretation of results regarding incongruent scenes. Using N2 posterior-contralateral ERP component as a primary measure, we revealed that threatening images indeed capture attention automatically and rapidly, but semantically incongruent scenes do not benefit from an automatic attentional selection. Thus, our results suggest that identification of the context-object mismatch is not preattentive.

Facilitation of allocentric coding by virtue of object-semantics

In the field of spatial coding it is well established that we mentally represent objects for action not only relative to ourselves, egocentrically, but also relative to other objects (landmarks), allocentrically. Several factors facilitate allocentric coding, for example, when objects are task-relevant or constitute stable and reliable spatial configurations. What is unknown, however, is how object-semantics facilitate the formation of these spatial configurations and thus allocentric coding. Here we demonstrate that (i) we can quantify the semantic similarity of objects and that (ii) semantically similar objects can serve as a cluster of landmarks that are allocentrically coded. Participants arranged a set of objects based on their semantic similarity. These arrangements were then entered into a similarity analysis. Based on the results, we created two semantic classes of objects, natural and man-made, that we used in a virtual reality experiment. Participants were asked to perform memory-guided reaching movements toward the initial position of a target object in a scene while either semantically congruent or incongruent landmarks were shifted. We found that the reaching endpoints systematically deviated in the direction of landmark shift. Importantly, this effect was stronger for shifts of semantically congruent landmarks. Our findings suggest that object-semantics facilitate allocentric coding by creating stable spatial configurations.