MétaCan
Menu
Back to cohort
Record W2752713396 · doi:10.1167/17.10.504

Ruling out task difficulty in the context-generalization of texture perceptual learning

2017· article· en· W2752713396 on OpenAlexaff
Alicia Campos Serrano, Ali Hashemi, Allison B. Sekuler, Patrick Bennett

Bibliographic record

VenueJournal of Vision · 2017
Typearticle
Languageen
FieldComputer Science
TopicDomain Adaptation and Few-Shot Learning
Canadian institutionsMcMaster University
Fundersnot available
KeywordsArtificial intelligencePerceptionContext effectContext (archaeology)Computer visionGeneralizationComputer sciencePerceptual learningPattern recognition (psychology)Cognitive psychologyPsychologyMathematicsGeometryNeuroscienceGeography

Abstract

fetched live from OpenAlex

Perceptual learning in a texture identification task reflects improved sensitivity to diagnostic features. We have studied perceptual learning using orientation-filtered textures that contain identity-specific information in a horizontal orientation band (i.e., Target) and non-diagnostic information (i.e., Context) in a vertical orientation band. Using Target-alone or Target+Context textures as training stimuli in a 1-of-6 texture identification task, Hashemi et al. (VSS 2015) demonstrated that learning was significantly more difficult in the Target+Context condition. Interestingly, the more difficult condition produced greater context generalization (Hashemi et al., VSS 2016): Target+Context trained observers generalized learning to the same Targets alone (i.e., presented without Context), but Target-alone trained observers did not transfer their overall strong learning to the same Targets with Context. These results may reflect a difference in perceptual strategies: Target-alone textures can be identified using any visible pixel, while Target+Context textures require observers to learn the selective extraction of target information in a specific orientation band and ignore the context. However, identifying Target+Context stimuli was significantly more difficult than Target-alone textures, so it is possible that the asymmetry in context generalization is a by-product of task difficulty and/or the magnitude of learning during training. Here, we tested that idea by adjusting stimulus contrast to make identification of Target-alone textures as difficult as Target+Context textures. We found that equating task difficulty did not eliminate the difference in context generalization: Target+Context training improved identification of both Target+Context and Target-alone textures, but Target-alone training improved identification only of Target-alone textures. We conclude that context-generalizable learning reflects a perceptual strategy learned when observers have to distinguish diagnostic from non-diagnostic information, and is not simply a by-product of task difficulty. Meeting abstract presented at VSS 2017

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Full frame distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.001
metaresearch head score (Gemma)0.000
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Observational · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.681
Threshold uncertainty score0.240

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0010.000
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0000.000
Science and technology studies0.0000.000
Scholarly communication0.0000.001
Open science0.0010.000
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0000.000

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.028
GPT teacher head0.311
Teacher spread0.283 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one teacher head, not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designObservational
Domainnot available
GenreEmpirical

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations0
Published2017
Admission routes1
Has abstractyes

Explore more

Same venueJournal of VisionSame topicDomain Adaptation and Few-Shot LearningFrench-language works237,207