Subregions in the ventromedial prefrontal cortex integrate threat and protective information to meta-represent safety

Copyright and License

© 2025 Tashjian et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding

The study was supported by the following grants to SMT: US National Science Foundation Directorate for Social, Behavioral and Economic Sciences Postdoctoral Research Fellowship 2203522 (https://www.nsf.gov/); Brain and Behavior Research Foundation Young Investigator Grant 30788 (https://bbrfoundation.org/). The study was supported by the following grants to DM: US National Institute of Mental Health grant R01MH133730 (https://www.nimh.nih.gov/); John Templeton Foundation grant TWCF0366 (https://www.templeton.org/). The funders played no role in study design, data collection or analysis, decision to publish, or preparation of the manuscript.

Data Availability

Task code and behavioral data are available through OSF, https://osf.io/hw3r9, DOI 10.17605/OSF.IO/8QG7Y. Neuroimaging data are available through Science Data Bank, https://www.scidb.cn/en/s/BNjaUz, DOI 10.17605/OSF.IO/8QG7Y.

Supplemental Material

S1 Text. Table A in S1 Text.

Mixed effects models predicting safety prediction (probability of choosing “win”) from stimuli safety value, split by stimulus order and type. Table B in S1 Text. Neural response to safety. Significant clusters from group level whole-brain univariate analyses. Table C in S1 Text. Neural response to danger. Significant clusters from group level whole-brain univariate analyses. Fig A in S1 Text. Figure A. Neural response to danger increases during each task phase, highlighting regions of canonical defensive circuitry involved in threat detection such as the insula, thalamus, and PAG. All analyses were conducted using FSL Randomise, TFCE, FWE-corrected p < 0.05. Color bar indicates t-intensity values. (Fig A panels A–C) Parametric increases in whole-brain neural activity that track decrease in experimentally established safety value of stimuli during Danger Prediction. The first stimulus presented represented a bias to partial information, which measures a differentiation in neural activity as a function of stimulus type (threat versus protection). Significant clusters indicate activation increased in those regions as safety probability decreased. Safety decrease was based on the average experimentally established safety probability of each stimulus (protection continuum order: fist, stick, gun grenade; threat continuum order: cat, goose, lion, grizzly). (Fig A panel A) Threat and Protection collapsed, (Fig A panel B) Threat only, (Fig A panel C) Protection only. (Fig A panels D–F) Parametric increases in whole-brain neural activity that track the increased experimentally established safety value of stimuli during Danger Meta-representation. The second stimulus safety value was based on the combined danger probability of the first and second stimuli. For analyses, safety was based on comparison with the average safety value of the stimulus and examined for trials where safety decreased. For example, if a stick was shown as the second stimulus and was paired with a lion, the probability of safety would reduce from 35.72% (safety average for all stick trials) to 21.43% (safety when stick is paired with lion) (see Fig 1B). (Fig A panel D) Threat and Protection collapsed, (Fig A panel E) Threat only, (Fig A panel F) Protection only. (Fig A panel G) Neural activation in response to Danger Recognition when subjects learned they were unsuccessful in battle. Analyses probed response at the outcome screen when it indicated potential for electric shock (20%) compared to when it indicated certain safety from shock (100%). Source data can be found at https://osf.io/8qg7y/ under “MRI data.” Fig B in S1 Text. Results of stimuli development and selection of stimuli at the high and low ends of the safety estimation spectrum. Two questions were asked related to level of danger (animals) and power (weapons) of 20 potential stimuli images. Items were also paired in head-to-head battles with all other stimuli of the same type. Lion and grizzly were rated as the most dangerous stimuli and cat and goose were rated as the second and third least dangerous stimuli (rat was selected as the least dangerous but ultimately excluded from the set to avoid conflating threat with disgust). The same rankings were reported for the head-to-head battles across all animals. The grenade and gun were rated as the most powerful weapons and as most likely to win head-to-head. Fist and stick were rated in the bottom 30% of power ratings and bottom 20% of head-to-heads. Other weapons rated as less powerful were excluded due to concerns of unwieldy usage (i.e., rope). Stimulus design was inspired by 2021 YouGov survey of 1,224 adults showing that that 6% of Americans believe they could beat a grizzly bear in a fight without weapons (retrieved from: https://today.yougov.com/society/articles/35852-lions-and-tigers-and-bears-what-animal-would-win-f). Animals were selected as pilot stimuli based on those survey results and a selection of weapons across the range of potential danger was also tested. Animal images were selected to be “attacking” to mitigate any issues of liking the animal (cuddly housecat versus angry housecat) to ensure stimuli were immediately recognizable as a threat. Weapon images were selected to be without any other stimuli in the picture and on white backgrounds. All images were kept in black and white and sepia tone ranges.

https://doi.org/10.1371/journal.pbio.3002986.s001 (DOCX)