United States

We propose and implement a relatively simple computational neural-network model of number comparison. Training on paired comparisons of the integers 1-9 enables the model to efficiently and accurately simulate some fundamental empirical phenomena (distance and ratio effects on accuracy and response time). It also generalizes robustly to more advanced tasks involving multidigit integers, negative numbers, and decimal numbers. The work demonstrates that small neural networks can sometimes efficiently learn a powerful system that exhibits extremely robust generalization to untrained items. Some important alternate models of number comparison are considered to establish a broader context. Several predictions and suggestions are made for future empirical and computational research in this area.

CogSci 2023

A Neural Model of Number Comparison with Robust Generalization

poster

The 45th Annual Meeting of the Cognitive Science Society will take place in Sydney, Australia as a modified hybrid conference. Presentations from the main-session room will be live-streamed every day for all events. This will include all invited talks and award sessions. Recordings of all CogSci 2023 sessions will be made available for on-demand viewing after the conference.<br>
The in-person component will be held at the International Convention Centre Sydney at Darling Harbour. The Conference will take place Wednesday July 26th – Saturday July 29th, 2023 (Wednesday, July 26th is the Workshops Day).

Join us!

<br>[**The Program at a Glance can be found here**](https://cognitivesciencesociety.org/wp-content/uploads/2023/07/COgSci-PAAG-V2.pdf)
<br>[**The Detailed Program can be found here**](https://online.fliphtml5.com/ebtyf/mswz/)
<br><br>

#### **CogSci 2023 Program Book**
<iframe style="width:700px;height:400px" src="https://online.fliphtml5.com/ebtyf/tzgk/"  seamless="seamless" scrolling="no" frameborder="0" allowtransparency="true" allowfullscreen="true" ></iframe>
<br>

The 45th Annual Meeting of the Cognitive Science Society presents the latest research across cognitive science and highlights the theme of Cognition in Context.

Auditory perception requires categorizing sound sequences, such as speech, into classes, such as syllables. Such categorization depends not only on the sequences’ acoustic waveform, but also on the listener’s sensory uncertainty, any individual sound’s relevance to the task, and learning the temporal statistics of the acoustic environment. Although previous studies have explored the effects of these perceptual and cognitive factors in separation, whether and how their interplay shapes categorization is unknown. Here, we tested this interplay by measuring human participants’ performance on a multi-tone categorization task. Using a Bayesian framework, we found that task-relevant tones contributed more to category choice than task-irrelevant tones, confirming that participants combined information about sensory features with task relevance. Conversely, poor estimates of tones’ task relevance or high sensory uncertainty adversely impacted category choice. Learning temporal statistics of sound category also affected decisions – the magnitude of this effect correlated inversely with participants’ relevance estimates. These results demonstrate that humans differentially weigh sensory uncertainty, task relevance and statistical learning, providing a novel understanding of sensory decision-making under real-life behavioral demands.

The Interplay of Relevance, Sensory Uncertainty and Statistical Learning Influences Auditory Categorization

Prior work has demonstrated the presence of hysteresis effects in the control of affordance-guided behavior, in that behavioral transitions around a critical action boundary vary with directions of change in said action boundary. To date, research on this topic has overlooked the influence of the global context on these phenomena. We employ an affordance-based reaching task, whereby participants were asked to move a target to a goal by passing through one of two apertures (size variable or size constant). It was found that the direction of change in the size of the variable aperture influenced the point of behavioral transitions, and this effect interacted with the location of a proceeding goal. In addition, we considered fluctuations in the entropy of participants’ reach trajectories as a window into the nature of the behavioral phase transitions. Differences in the structure of entropy were found depending on the direction of change in the size variable aperture. These results are discussed considering a dynamical systems approach, and recommendations for future work are made. 

Keywords: affordances, decision making, self-organization, behavioral transitions

Moving the Goalposts: The Influence of Context on Behavioral Transitions in a Unilateral Manual Reaching Task

Social media use has changed how people relate to one another in fundamental ways. Amongst other innovations, the widespread use of image filters has allowed for people to change how they want to be perceived by other users by altering the shape of their bodies to fit current beauty ideals. Research on this area has focused primarily on the social factors that shape filter use in adolescent women. This study investigates the early cognitive processing of filtered images; mainly whether young adults can detect the presence of filters that produce large (Experiment 1) and subtle (Experiment 2) body modifications to either approach of distance it from beauty ideals prevalent for each gender. Results showed that both men and women can readily detect the presence of a filter, and they are faster and more accurate in responding to stimuli of their own gender compared to the opposite gender.

The Influence of Beauty Ideals on Instagram Filter Detection in Young Adults

The brain represents acoustic structure in speech and music hierarchically, with lower-order regions processing shorter-timescale information (e.g., syllables, notes) and higher-order regions processing longer timescales (e.g., sentences, musical phrases). However, it is not known to what extent this neural hierarchy reflects differences in cognitive processing of these distinct timescales. In a behavioral experiment, musician and non-musician participants heard naturalistic piano music scrambled at four temporal levels: 1-measure, 2-measure, 8-measure, and fully intact. We found that participants more accurately remembered and predicted musical information within a more intact musical context than a less intact one (p < .001). Stimuli had a consistent timbre and tempo, indicating that listeners were sensitive to tonal and rhythmic structure, which likely boosted their processing of the more intact, cohesive stimuli. Highly-trained musicians also outperformed non-musicians only in the memory task (p < .01), which suggests that implicit knowledge may be more useful for prediction.

Disrupting naturalistic temporal structure impacts musical memory, prediction, and segmentation

In assessing the sense of agency (SoA) in the field, researchers are often confronted with the limitations of conventional methods (e.g., retrospectivity, task-disruption). Generally, facial analysis offers a promising possibility to overcome these obstacles by directly collecting objective data in real-time and without any need for contact. This paper examines the applicability of face tracking for SoA assessment by investigating whether facial expressions can be used to derive information about the extent to which users feel in control in interactions with technical systems. In a laboratory experiment (N = 27) based on the dot control task, we manipulated SoA via task difficulty and explored the effects on facial expression patterns. Three action units in the region of the mouth were observed to be characteristic facial indicators of SoA. These findings support the suitability of facial expressions for SoA evaluation as well as the potential of face tracking for monitoring SoA in various practical fields of application (e.g., automated driving).

Are Facial Expressions Predictors for the Sense of Agency in a Dot Control Task?

Multimodal social signals are used when a third person perceives the relationship between multiple people. In this study, in order to determine whether people can detect improvisation played by skilled actors from audio-visual information, improvised and scripted play stimuli were presented online and participants were asked to identify which was the improvised play. Participants (n＝216) were randomly divided into three conditions (Audiovisual, Visual only, and Auditory only) and asked about their confidence level. Results showed significantly higher rate of correct response in Audiovisual and Auditory-only conditions than in Visual-only condition. In addition, examination of the free-text responses revealed that those who answered correctly in the audiovisual condition relied primarily on pauses in conversation to identify. These results suggest that in the performances of skilled actors, an improvisation is manifested in auditory signals as social information, which can be detected from the third person.

An on-line study about recognition of improvisation theatre using audio-visual information

We cannot help but form a first impression every time we meet someone unfamiliar to us. While the existing literature has almost exclusively focused on facial impressions, we know from our everyday experiences that other aspects of a person can also inform the judgements we make. Here, I use a data-driven approach to reveal the underlying structure of first impressions attributed to human voices, first names and those based on multimodal information (face, voice and name). Approximately 40,000 spontaneous descriptors were used to establish the most common first impression traits and Exploratory Factor Analyses revealed a number of fundamental dimensions, capturing judgements of approachability, confidence, and intelligence. Surprisingly, multimodal first impressions were primarily driven by voice cues and this was observed even when the presentation order of the three cues (faces, voices, names) was systematically manipulated, providing evidence for the remarkable role of voices in the formation of first impressions.

Multimodal Person Evaluation: First Impressions from Faces, Voices, and First Names

In this study, we investigate the priming effect of camera-induced objectifying-gaze, operationalised via sexualised music video(MV), on university students when looking at target female images in traditional (saree, salwar) and western (shirt-pant, short-dress) attires in the Indian demographics. We showed participants two videos where female leads wore unrevealing Indian attire (saree) and revealing western clothing in sexualised and traditional MVs respectively. We quantified gaze as fixation-duration and visit-count (revisit frequency) metrics. Our findings report that there is a priming effect of sexualised media. Female images in all attires are equally objectified and body-biased gaze is more pronounced in sexualised MV priming, short-dress is more objectified in traditional MV priming, and saree is less objectified overall, which confirms the social stereotypes on attire and respectability — traditionally clad (vs. non-traditional) women are seen as more moral and worthy of respect and protection compared to women in hypersexualised and objectified roles.

The Objectifying Gaze: Impact of Sexualized Media on Viewer Gaze Behavior towards (Non)Traditional Attire

Combining perceptual learning techniques with adaptive learning algorithms has been shown to accelerate the development of expertise in medical and STEM learning domains. Virtually all adaptive learning systems have relied on simple accuracy data that does not take into account response bias, a problem that may be especially consequential in multi-category perceptual classifications. We investigated whether adaptive perceptual learning in skin cancer screening can be enhanced by incorporating signal detection theory (SDT) methods that separate sensitivity from criterion. SDT-style concepts were used to alter sequencing, and separately to define mastery (category retirement). SDT retirement used a running d’ estimate calculated from a recent window of trials based on hit and false alarm rates. We used a Skin Cancer PALM (perceptual adaptive learning module) to train classification of 10 cancerous and non-cancerous skin lesion types. Four adaptive conditions varied either the type of adaptive sequencing (standard vs. SDT) or retirement criteria (standard vs. SDT). A control condition presented didactic instruction on dermatologic screening in video form, including images, standard classification schemes, and detailed explanations. All adaptive conditions robustly outperformed the control condition in both learning efficiency and fluency (large effect sizes). Use of SDT retirement criteria produced greater learning efficiency at both immediate and delayed posttests (medium effect sizes). SDT and standard adaptive sequencing did not differ. SDT enhancements to adaptive perceptual learning procedures have potential to enhance learning efficiency.

Connecting Adaptive Perceptual Learning and Signal Detection Theory in Skin Cancer Screening

Scatterplot research has identified factors that impact people’s perception of correlation magnitudes, yet much less is known about how people reason about data represented in scatterplots. We investigated how people make correlational and causal inferences based on scatterplots with and without outliers. In Experiment 1 and 2, participants viewed scatterplots matched in overall correlational magnitude depicted, but half had an outlier. In Experiment 3, the scatterplots in the two conditions were matched in the correlation magnitude depicted by all the dots excluding the outlier. For each scatterplot, participants stated their endorsement for correlational (X and Y change together) and causal statements (X changes Y). Only when outliers further strengthened an already moderate to strong relationship, people endorsed related correlational statements more and showed a stronger causality bias. Altogether we demonstrate that the impact of outliers in scatterplots on visual reasoning depends on the strength of the relationship depicted.

Next from CogSci 2023

The Interplay of Relevance, Sensory Uncertainty and Statistical Learning Influences Auditory Categorization

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES