**Probabilistic Learning of Emotion Categories**
*in press, Journal of Experimental Psychology: General*
Rista Plate\*, Adrienne Wood\*, Kristina Woodard, & Seth Pollak
*Contributed equally
University of Wisconsin - Madison
Abstract:
Although the configurations of facial muscles that humans perceive vary continuously, we represent emotion expressions as categories. This suggests that, as in other areas of categorical perception, humans become attuned to perceptual features of emotion cues and determine meaningful thresholds for these signals given the environment. However, little is known about how individuals come to learn and represent these salient social signals. In Experiment 1 we probe the contribution of supervised and unsupervised learning to the acquisition and maintenance of emotion categories in both children (6-8-years-old) and adults (18-22-years-old). Children and adults learned the boundary between neutral and angry when provided with explicit feedback (supervised learning). However, participants rapidly shifted category boundaries after unsupervised exposure to different statistical distributions. In Experiments 2 and 3, we replicated this finding and also tested whether adults are able to track statistical distributions for multiple actors at once. Not only did participants form actor-specific categories, but the distributions of facial features also influenced participants’ trait judgments about the actors. Taken together, these data are consistent with the view that the way humans construe emotions is not only flexible, but reflects complex learning about the distributions of cues individuals experience in their environments.
**NOTES ON STUDY 1 FILES**
* Data file is named "CAT_processed_combined.csv"
* Codebook is named "codebook_for_CAT_processed_combined_datafile.xlsx"
* R script for data analyses is named "CAT\_logistic\_regression.r"
* R markdown file (with complete R code and output) is named "cat1\_data\_analyses\_OSF.html"
* All non-face images used in the task are available in the images folder. Contact us (adrienne.wood@wisc.edu) for the face images, which we are not allowed to post for copyright reasons.
* Zip file containing incomplete materials for running Psychopy experiment is called "taskfiles.zip". **For image copyright reasons, please contact us (adrienne.wood@wisc.edu) for the images of Jane, then put them in the image folder. To run the experiment, first download and install Psychopy2.**
* Supplementary analyses are in the document "Study 1 Supplementary Analysis.docx". In these analyses, we ask whether participants' new category boundaries actually reached the midpoint of the distributions they were observing in the unsupervised learning phase.
**NOTES ON STUDIES 2A and 2B FILES**
* **We renamed the studies Experiment 2 and Experiment 3 for the manuscript, which is not reflected in all the files here.**
* Main data file for Study 2A is named "Cat_multAct_long.csv"
* Codebook is named XXXXXXXXX
* R Markdown script for data analyses is named "CAT_mult_actors_analyses.Rmd"
* R Markdown knitted file with code, output, and figures, is named "CAT_mult_actors_analyses.html"
* All non-face images used in the task are available in the images folder. Contact us (adrienne.wood@wisc.edu) for the face images, which we are not allowed to post for copyright reasons.
* The zip file contains the incomplete materials for running the multiple actors versions of the Psychopy multiple actors task, Study 2A. **For image copyright reasons, please contact us (adrienne.wood@wisc.edu) for the images of Jane, then put them in the image folder. To run the experiment, first download and install Psychopy2.**
* Supplementary analyses are in the document "Studies 2 and 3 Supplementary Analyses.docx". In these analyses, we ask whether participants' new category boundaries actually reached the midpoint of the distributions they were observing in the unsupervised learning phase.
