Jonathan Doriscar
Northwestern PhD candidate · M.S. Statistics and Data Science · NSF GRFP
I build intelligent systems for human complexity.
I'm Jonathan Doriscar — a cognitive scientist, computational behavioral scientist, and founder working across memory, language, behavior, data, and AI.
I run experiments, model messy behavioral data, and analyze language at scale to study how people make sense of complicated information.
Currently finishing my dissertation and building MemwaMind.
How I think — focus, methods, public work, and what I’m building.
A compact console listing my focus, methods, public work, and current builder direction.
> focus cognition | data | language | judgment > methods experiments | causal inference | multilevel modeling | NLP & LLMs | applied AI > public work Social Cognition | IPR working paper | Annual Review of Psychology | JPSP | Nature Energy > building MemwaMind: memory, documents, review, professional work
focus
cognition | data | language | judgment
methods
experiments | causal inference | multilevel modeling | NLP & LLMs | applied AI
public work
Social Cognition | IPR working paper | Annual Review of Psychology | JPSP | Nature Energy
building
MemwaMind: memory, documents, review, professional work
Technical proof
Methods I use in public and product work.
Most of what I do runs on the same set of tools. I design experiments and model messy behavioral data, analyze language at scale, train and evaluate machine learning models, and build AI workflows that keep evidence and human review close at hand.
Northwestern IT
Research computing and data science profile
Institutional profile connecting my applied data science work — R, Python, natural language processing, machine learning — to Northwestern's research computing community.
Social Cognition
Unsupervised machine learning methods paper
Lead-author methods article using K-means, DBSCAN, PCA, and market basket analysis as discovery tools for social cognition.
IPR working paper
Why Reform Stalls
Two-study project with 257,401 public YouTube comments, LLM-assisted text analysis, and an experiment.
How I work
How I turn messy social data into interpretable questions.
I usually start with a messy human question, then pick the method that can make part of it testable — sometimes an experiment, sometimes a model, sometimes a text-analysis pipeline. The last step is always interpretation: what does the method actually show, and what still needs theory or human judgment? Each tab below is the real method I used on a real public project — not a stylized animation.
K-means clustering, K = 5
Cluster centers from Project Implicit data on Black-White Implicit Association Test (IAT) responses, N = 13,855.
Strongly liberal but with measurable pro-White / anti-Black implicit bias paired with warm explicit feelings — the classic explicit / implicit dissociation.
Data and code: From Data to Discovery (Doriscar et al., 2025, Social Cognition). Conceptual sketch only — I'm not reproducing model performance or any figure from the paper here.
From social attitudes to interpretable structure
large-scale attitude data + Project Implicit Race IAT worked examples + responses and items that may contain hidden structure can become possible clusters, dimensions, and co-occurring response patterns that researchers can interpret with theory.
- Topic
- Social attitudes and belief patterns
- Method
- Unsupervised machine learning
- Public project
- From Data to Discovery - Published in Social Cognition
- Takeaway
- Where I bridge cognitive science, machine learning, and methods education — without letting the model replace interpretation.
- What this does not show
- Conceptual sketch only — I'm not reproducing model performance or any figure from the paper here.
Selected projects
Projects across research, methods, and applied AI.
These all started the same way — with a real question I couldn’t shake, and a method that actually fit the material. On each page I try to stay honest about what came out and what didn’t.
Unsupervised Machine Learning for Social Cognition
Using unsupervised machine learning to reveal hidden patterns in human beliefs and attitudes.
Role: Lead author.
Why Reform Stalls
Modeling public justification, outrage, and reform discourse around police violence.
Role: Conceived and led the project; conceptualization, methodology, data curation, formal analysis, validation, investigation, visualization, and writing.
Historical Blame and Collective Responsibility
Understanding why people hold present-day groups responsible for past harms.
Role: Coauthor.
Founder project
I’m building MemwaMind.
A workspace for professional memory and review in tax and accounting work.
Document orientation: every claim stays connected to the material behind it.
Professional workflow: drafts and human review, not unbounded automation.
Institutional context
Selected profiles, coverage, and publications.
A few public pages where my training, research, and institutional context show up externally.
New Tools for Studying Bias and Belief
IPR coverage of the Social Cognition methods article.
IPR coverage of my Social Cognition article on unsupervised machine learning methods for social cognition.
Open page
Why Reform Stalls: Justification and Outrage as Competing Public Responses to Police Violence
IPR working paper source page, WP-25-31.
Public IPR working paper page for my two-study project on justification, outrage, and reform discourse around police violence.
Open page
TGS Spotlight: Jonathan Doriscar
Northwestern's Graduate School spotlight page.
Northwestern Graduate School profile of my work as a PhD candidate in Social Psychology and a Master's student in Data Science & Statistics.
Open page
Publications.
From Data to Discovery: Unsupervised Machine Learning’s Role in Social Cognition
Social Cognition, 2025
Why Reform Stalls: Justification and Outrage as Competing Public Responses to Police Violence
Northwestern Institute for Policy Research Working Paper Series, WP-25-31, 2025
When the Specter of the Past Haunts Current Groups: Psychological Antecedents of Historical Blame
Journal of Personality and Social Psychology, 2024
Assessing How Energy Companies Negotiate with Landowners When Obtaining Land for Hydraulic Fracturing
Nature Energy, 2024
Funding and honors.
National Science Foundation Graduate Research Fellowship
Competitive graduate fellowship supporting my research training across cognitive science, computational methods, and applied AI.
Edward Bouchet Graduate Honor Society Scholar
Honor recognizing scholarly achievement and broader commitments in graduate education.
Contact
Get in touch.
Whether it’s a research collaboration, an applied AI or machine learning question, or a product conversation, I’d be glad to hear from you.