20 papers · ranked by Valyu relevance
Pritam Kadasi, Abhishek Upperwal, Mayank Singh
We propose the Task–Specificity Score (TSS) to quantify how much an instruction matters for predicting its output, by contrasting the true instruction against plausible alternatives for the same input. We further introduce TSS++, which uses hard alternatives and a small quality term to mitigate easynegative effects.…
Kim, Olivia
Prompt design plays a critical role in the reasoning performance of large language models (LLMs), yet the impact of prompt specificity—how detailed or vague a prompt is—remains understudied. This paper introduces DETAIL, a framework for evaluating LLM performance across varying levels of prompt specificity. We generate…
Michael Li, Nishant Subramani
The circuits framework in mechanistic interpretability aims to identify causally important sparse subgraphs of model components, typically evaluated by measuring necessity and sufficiency. We measure circuit reuse, the proportion of components shared across per-example circuits within a task, and investigate two…
Joshua B. Tan, Isabella F. Orlando, Jungwoo Kim, Christopher J. Cueva + 6 more
Human cognition depends on the ability to flexibly recombine existing knowledge in new ways. Although this capacity for compositionality has traditionally been attributed to cortical networks, its broader neural basis remains unclear. Here, we combined dimensionality reduction of task-based fMRI with recurrent neural…
Giorgio L. Manenti, Caspar M. Schwiedrzik
Perceptual learning improves sensory discrimination, yet the brain must balance specificity and generalization, especially in variable environments. To investigate how variability shapes perceptual learning, we used functional magnetic resonance imaging while human subjects performed an orientation discrimination task…
Matthew K. Robison, Stephen Campbell, Lauren D. Garner, Ciara Sibley + 1 more
The present study examined individual differences in 24 measures of cognitive ability in a sample of young adults (N = 255). Each measure was completed twice, separated by a period of 2 weeks, to assess test-retest reliability and retesting (i.e., practice) effects. Latent variable modeling was used to assess the…
Authors not listed
A framework for catalysis based on categorical aperture selection rather than temporal acceleration is presented. Traditional catalysis theory describes catalysts as agents that accelerate reactions by lowering activation energies, implicitly treating time as the fundamental variable and reaction rate enhancement as…
Rhea Kapur, Robert D. Hawkins, Elisa Kreiss
Vision-language models (VLMs) are increasingly used to make visual content accessible via text-based descriptions. In current systems, however, description specificity is often conflated with their length. We argue that these two concepts must be disentangled: descriptions can be concise yet dense with information, or…
Sina Tafazoli, Flora M. Bouchacourt, Adel Ardalan, Nikola T. Markov + 4 more
Cognition is highly flexible-we perform many different tasks1 and continually adapt our behaviour to changing demands2,3. Artificial neural networks trained to perform multiple tasks will reuse representations4 and computational components5 across tasks. By composing tasks from these subcomponents, an agent can…
Simon Leipold, Ryssa Moffat
Studying learning-related plasticity is central to understanding the acquisition of complex skills, for example learning to master a musical instrument. Over the past three decades, conventional group-based functional magnetic resonance imaging (fMRI) studies have advanced our understanding of how humans' neural…
Deanna L. Strayer, Nash Unsworth
Attention lapses occur when focus shifts away from the task at hand towards internal or external distractions and can lead to failures in completing intended actions. Goal-setting theory proposes that setting specific, difficult goals leads to better task performance over vague goals. The present study examined whether…
Mengqiao Chai, Iris Ikink, Stefania Mattioni, Ricardo Alejandro Benavides + 5 more
People can dynamically and adaptively update task goals in the face of task uncertainty. In this fMRI study, we aimed to investigate the modulation of neural task representations that subserve such flexible task control. On each trial, we asked people to perform one of nine image categorization tasks. During task…
Christopher R. Nolan, Mike E. Le Pelley, Kelly G. Garner
The benefits of routines for daily functioning are widely acknowledged, yet, despite their apparent importance, methods for quantifying routine maintenance and the causes of their disruption remain lacking. Here, we propose a novel means of defining and quantifying routines (transition entropy). Using the transition…
Rick den Otter, Anna Dame, Sjoerd Stuit, Leendert van Maanen + 1 more
Theories of dual-task interference assume that the same cognitive operations underlie multitasking regardless of stimulus timing, yet this core assumption has remained untested due to methodological limitations of behavioral averaging. Here, we combine hidden multivariate pattern (HMP) analysis with deep spatiotemporal…
David A. Neequaye, Alexandra Lorson, Holly K. Barnett
We examine the mechanisms by which interviewees in investigative interviews mentally organize information when deciphering what an interviewer wants to know. The overarching idea is that such a process stems from the extent to which an interviewer’s question specifies an objective. Our initial test (i.e., [15])…
Evgeniia Alenina, Kristina Terenteva, Vladimir Kosonogov
Anxiety, characterized by pervasive feelings of nervousness and worry, may differentially impact cognitive problem-solving across various domains, including spatial and social contexts. This study investigates the relationship between distinct types of anxiety and cognitive performance, aiming to elucidate the…
Authors not listed
Incorporating prior domain knowledge into Bayesian optimization (BO) remains difficult for statistical methods, which also typically suffer from limited interpretability. Large language models (LLMs) offer complementary strengths in reasoning and knowledge integration, but it remains unclear when and how they improve…
Authors not listed
Machine olfaction—the artificial replication of the sense of smell—faces significant challenges due to the absence of large, standardized training datasets. Unlike vision, language, and audio models, which benefit from extensive corpora such as ImageNet, GLUE, and AudioSet, olfaction lacks scaled equivalents and…
Martijn P. van den Heuvel, Ilan Libedinsky, Sebastian Quiroz Monnens, Jonathan Repple + 1 more
Lesion Network Mapping (LNM) is a framework used for identifying symptom-related brain circuits by projecting lesion locations onto a normative connectome. Recent methodological investigations have raised concerns about the biological interpretation and specificity of the circuits derived using this method, with…
Leslie K. Held, Judith Goris, Senne Braem
In our daily lives, we often need to switch between states of cognitive flexibility and stability based on environmental demands. It has been suggested that people can learn to navigate this balance based on reinforcement, but that this may be impaired in autism or other transdiagnostic dimensions. Replicating previous…