About this site

This is the website of the Cognitive and Affective Neuroscience Lab (CANlab) at Dartmouth College, directed by Tor Wager. It replaces the lab's earlier site at sites.dartmouth.edu/canlab and keeps all of its content: research themes, people, the full publication list with PDFs and resources, news, training materials, and how to join or take part in a study. It adds a searchable, tagged publication database, reading lists and reference export, a collaborator map, and an automated daily search for new papers. The whole site is a static site built from plain JSON data in one public GitHub repository, following the design of the lab's Science of Placebo research hub.

The publication database and its tags

• Records. Every paper, preprint and chapter with Tor Wager as an author, from the lab's publication list and CV; editorials and commentaries written about a paper by others are attached to that paper rather than listed separately. Records come with the original links (PDFs, publisher pages, brain pattern maps, paradigms, code, data and videos). Records were matched to OpenAlex to add DOIs, PubMed IDs, abstracts, full author lists and citation counts.
• Tags. Each paper carries keywords on three axes: topic (placebo effects, pain, chronic pain, emotion, emotion regulation, empathy, stress, expectation and learning, social influence, reward and craving, cognitive control, brain-body interactions, clinical, perception), approach (fMRI, neuroimaging methods, neuromarker, machine learning, AI integration, computational model, mediation, meta-analysis, mega-analysis, pharmacology, psychophysiology and EEG, brain stimulation, other imaging, behavioural, open tools) and article type. Tags were assigned by a large language model that read the full text of each paper where a PDF is available (the methods and results, not just the abstract, so that features such as a deep-learning analysis or a mediation model are tagged even when the abstract does not mention them), and from the abstract otherwise. Each paper page states which text was used and the model's confidence.
• Corrections. Machine tags will sometimes be wrong. Every paper page has a link to suggest a correction; corrections are applied by hand in the data file and override the model.
• PDFs. Every paper links to a PDF stored with the site (the publisher's version wherever one could be obtained; PubMed Central author manuscripts are labelled as such). Publisher PDFs are provided for personal, scholarly use.

Daily search for new papers and news

• Every day a workflow queries OpenAlex and PubMed for new papers by the lab, tags candidates with the same model and vocabulary, and adds them to a review queue shown on the publications page. Approving a candidate (a one-line edit in the data file, or a click for signed-in lab members once the community API is deployed) moves it into the database and the RSS feed.
• Every two days a second workflow collects news stories and podcast episodes mentioning the lab and its work and lists them under News.
• Citation counts are refreshed weekly from OpenAlex.

Reading lists, export and accounts

Star papers to keep them in My list, save the selection under a name to build several lists, and export any list as BibTeX, RIS, CSV, a DOI list or formatted citations. Lists live in your browser; signing in with GitHub syncs them across devices through a small community API (a Cloudflare Worker) that stores only your GitHub id, login and avatar together with your lists.

Privacy and analytics

The site sets no cookies of its own. Traffic is measured with cookie-free, privacy-preserving analytics that record page views without identifying visitors. Sign-in sessions are random tokens in an HttpOnly cookie on the API domain, valid for 30 days; signing out deletes the session.

Contributing

• Suggest a missing paper, a link to data or code, or a tag correction by opening an issue.
• Lab members: edit the JSON files under data/ in the repository (people, news posts, research themes, resources, openings); the site rebuilds automatically on push.
• The data files are plain JSON and may be reused with attribution. Code is MIT-licensed. Abstracts remain the copyright of their publishers and are reproduced for indexing and discovery.

Credits

Designed and built by Tor Wager with Claude Code. Fonts: Fraunces, Inter and JetBrains Mono. Search by MiniSearch; network and charts by d3. Bibliographic data from OpenAlex and PubMed.