skip to main content

Title: Macaques preferentially attend to intermediately surprising information.
Normative learning theories dictate that we should preferentially attend to informative sources, but only up to the point that our limited learning systems can process their content. Humans, including infants, show this predicted strategic deployment of attention. Here we demonstrate that rhesus monkeys, much like humans, attend to events of moderate surprisingness over both more and less surprising events. They do this in the absence of any specific goal or contingent reward, indicating that the behavioral pattern is spontaneous. We suggest this U-shaped attentional preference represents an evolutionarily preserved strategy for guiding intelligent organisms toward material that is maximally useful for learning.
Award ID(s):
Publication Date:
Journal Name:
Proceedings of the Cognitive Science Society
Sponsoring Org:
National Science Foundation
More Like this
  1. Moments of inattention to our surroundings may be essential to optimal cognitive functioning. Here, we investigated the hypothesis that humans spontaneously switch between two opposing attentional states during wakefulness—one in which we attend to the external environment (an “online” state) and one in which we disengage from the sensory environment to focus our attention internally (an “offline” state). We created a data-driven model of this proposed alternation between “online” and “offline” attentional states in humans, on a seconds-level timescale. Participants ( n = 34) completed a sustained attention to response task while undergoing simultaneous high-density EEG and pupillometry recording andmore »intermittently reporting on their subjective experience. “Online” and “offline” attentional states were initially defined using a cluster analysis applied to multimodal measures of (1) EEG spectral power, (2) pupil diameter, (3) RT, and (4) self-reported subjective experience. We then developed a classifier that labeled trials as belonging to the online or offline cluster with >95% accuracy, without requiring subjective experience data. This allowed us to classify all 5-sec trials in this manner, despite the fact that subjective experience was probed on only a small minority of trials. We report evidence of statistically discriminable “online” and “offline” states matching the hypothesized characteristics. Furthermore, the offline state strongly predicted memory retention for one of two verbal learning tasks encoded immediately prior. Together, these observations suggest that seconds-timescale alternation between online and offline states is a fundamental feature of wakefulness and that this may serve a memory processing function.« less
  2. Salient segmentation is a critical step in biomedical image analysis, aiming to cut out regions that are most interesting to humans. Recently, supervised methods have achieved promising results in biomedical areas, but they depend on annotated training data sets, which requires labor and proficiency in related background knowledge. In contrast, unsupervised learning makes data-driven decisions by obtaining insights directly from the data themselves. In this paper, we propose a completely unsupervised self-aware network based on pre-training and attentional backpropagation for biomedical salient segmentation, named as PUB-SalNet. Firstly, we aggregate a new biomedical data set from several simulated Cellular Electron Cryo-Tomographymore »(CECT) data sets featuring rich salient objects, different SNR settings, and various resolutions, which is called SalSeg-CECT. Based on the SalSeg-CECT data set, we then pre-train a model specially designed for biomedical tasks as a backbone module to initialize network parameters. Next, we present a U-SalNet network to learn to selectively attend to salient objects. It includes two types of attention modules to facilitate learning saliency through global contrast and local similarity. Lastly, we jointly refine the salient regions together with feature representations from U-SalNet, with the parameters updated by self-aware attentional backpropagation. We apply PUB-SalNet for analysis of 2D simulated and real images and achieve state-of-the-art performance on simulated biomedical data sets. Furthermore, our proposed PUB-SalNet can be easily extended to 3D images. The experimental results on the 2d and 3d data sets also demonstrate the generalization ability and robustness of our method.« less
  3. Goel, A ; Seifert, C ; Freska, C (Ed.)
    Humans often rely on past experiences stored in long-term memory to predict the outcome of an event. In traditional lab-based experiments (e.g., causal learning, probability learning, etc.), these observations are compressed into a successive series of learning trials. The rapid nature of this paradigm means that completing the task relies on working memory. In contrast, real-world events are typically spread out over longer periods of time, and therefore long-term memory must be used. We conducted a 24 day smartphone study to assess how well people can learn causal relationships in extended timeframes. Surprisingly, we found few differences in causal learningmore »when subjects observed events in a traditional rapid series of 24 trials as opposed to one trial per day for 24 days. Specifically, subjects were able to detect causality for generative and preventive datasets and also exhibited illusory correlations in both the short-term and long-term designs. We discuss theoretical implications of this work.« less
  4. Information manipulation is widespread in today’s media environment. Online networks have disrupted the gatekeeping role of traditional media by allowing various actors to influence the public agenda; they have also allowed automated accounts (or bots) to blend with human activity in the flow of information. Here, we assess the impact that bots had on the dissemination of content during two contentious political events that evolved in real time on social media. We focus on events of heightened political tension because they are particularly susceptible to information campaigns designed to mislead or exacerbate conflict. We compare the visibility of bots withmore »human accounts, verified accounts, and mainstream news outlets. Our analyses combine millions of posts from a popular microblogging platform with web-tracking data collected from two different countries and timeframes. We employ tools from network science, natural language processing, and machine learning to analyze the diffusion structure, the content of the messages diffused, and the actors behind those messages as the political events unfolded. We show that verified accounts are significantly more visible than unverified bots in the coverage of the events but also that bots attract more attention than human accounts. Our findings highlight that social media and the web are very different news ecosystems in terms of prevalent news sources and that both humans and bots contribute to generate discrepancy in news visibility with their activity.« less
  5. Present day ideals of good parenting are socio-technical constructs formed at the intersection of medical best practices, cultural norms, and technical innovation. These ideals take shape in relation to the fundamental uncertainty that parents/mothers face, an uncertainty that comes from not knowing how to do what is best for one's children, families, and selves. The growing body of parent-focused smart devices and data-tracking platforms emerging from this intersection frame the responsible parent as one who evaluates, analyzes, and mitigates data-defined risks for their children and family. As these devices and platforms proliferate, whether from respected medical institutions or commercial interests,more »they place new demands on families and add an implicit emphasis on how humans (often mothers) can be augmented and improved by data-rich technology. This is expressed both in the actions they support (e.g., breastfeeding, monitoring food intake), as well as in the emotions they render marginal (e.g., rage, struggle, loss, and regret). In this article, we turn away from optimization and self-improvement narratives to attend to our own felt experiences as mothers and designers. Through an embodied practice of creating Design Memoirs, we speak directly to the HCI community from our position as both users and subjects of optimized parenting tools. Our goal in this work is to bring nuance to a domain that is often rendered in simplistic terms or frames mothers as figures who could endlessly do more for the sake of their families. Our Design Memoirs emphasize the conflicting and often negative emotions we experienced while navigating these tools and medical systems. They depict our feelings of being at once powerful and powerless, expressing rage and love simultaneously, and struggling between expressing pride and humility. The Design Memoirs serve us in advocating that designers should use caution when considering a problem/solution focus to the experiences of parents. We conclude by reflecting on how our shared practice of making memoirs, as well as other approaches within feminist and queer theory, suggest strategies that trouble these optimization and improvement narratives. Overall, we present a case for designing for mothers who feel like they are just making do or falling short, in order to provide relief from the anxiety of constantly seeking improvement.« less