Ph.D. Candidate, Cognitive Science, IIT Kanpur

Tapas Ranjan Rath

Decision-making, computational modeling, human-AI interaction, and multi-agent alignment

Publications Download CV

About Me

I am a Ph.D. candidate in Cognitive Science at the Indian Institute of Technology (IIT) Kanpur. My doctoral research examines the computational mechanisms of context effects in human decision-making: how environmental structures such as irrelevant options, decoys, and attribute trade-offs bias choice, and how those deviations from rationality can be modeled robustly.

My work combines behavioral experimentation, simulation, Bayesian and mixed-effects modeling, and reinforcement-learning benchmarks. I also bring prior engineering experience from Accenture, where I worked on large-scale QA and software testing for British Telecom.

I am currently extending this work into human-AI interaction and AI alignment, especially questions about sense of agency, AI-mediated choice, value conflict, and how assistive or agentic systems can support human control rather than quietly displacing it.

GitHub IIT Kanpur Profile Research Statement

Methods & Tools

  • R: lme4, brms, BayesFactor
  • Python: Pandas, SciPy, PyTorch
  • Experimental design: PsychoPy, JavaScript
  • Eye tracking: Eyelink 1000 Plus

Research

Ph.D. Research

My doctoral work investigates context effects in decision-making, with a focus on how attribute structures such as decoys influence human choice. I build online and offline experiments in JavaScript and PsychoPy, and I use computational simulations to evaluate when common decision metrics remain valid under heterogeneous preferences. Code

I also model bounded rationality: how humans select adaptive strategies under resource constraints. Code

Current Projects: AI, Agency & Multi-Agent Alignment

  • Situated Agency Alignment (KARMA): Multi-agent RL project on proxy agency, plural values, role-invariant representations, cooperation, and harmful strategic drift in agent systems. Repository
  • DRQN for Probabilistic Reversal Learning: PyTorch model using reinforcement-learning agents as computational benchmarks for human adaptive decision-making and cognitive flexibility. Repository

Future Directions

  • Sense of agency in recommender systems: studying how algorithmic recommendations influence perceived control and decision ownership.
  • Sense of agency in agentic AI: examining how human control shifts when autonomous systems act on a user's behalf.
  • Value conflict in multi-agent AI: modeling how agents aligned to different users might coordinate, negotiate, and form norms under competing objectives.
  • Player agency in multiplayer games: understanding decision-making dynamics in complex, interactive digital environments.

Publications

Journal Articles

  • Rath, T., Srinivasan, N., & Srivastava, N. (2025). The attraction effect in perceptual decision-making: A case of dominance asymmetry. Frontiers in Psychology. Link

Selected Working Papers & Manuscripts in Preparation

Context Effects & Decision Architecture

  • Rath, T. (in final preparation). When decoys are easy to reject: Comparative asymmetry and context effects in consumer product-pack choice. Target journal: Journal of Consumer Psychology (Short Reports) Draft
  • Rath, T., Srivastava, N., & Srinivasan, N. (in final preparation). Pairwise comparison architecture in context-dependent choice. Working paper; target journal: Judgment and Decision Making Draft
  • Rath, T., Srivastava, N., & Srinivasan, N. (in final preparation). When difficult trade-offs make decoys work: Subjective equivalence geometry in the attraction effect. Target journal: Judgment and Decision Making Draft

AI Agency & Alignment

  • Rath, T. (in final preparation). The Proxy Agency Moral Shield: Extended Self, Semantic Fluency, and the Ethics of Agentic AI. Working paper; target journal: Minds and Machines Draft

Computational & Topological Modeling

  • Rath, T., Srivastava, N., & Srinivasan, N. (in final preparation). Unmasking the flaws of triplet-triplet attraction effect measures: Via mathematical analysis and agent-based simulations. Working paper; target journal: Journal of Mathematical Psychology Draft PsyArXiv
  • Srivastava, C., Rath, T., Beuria, J., & Gupta, R. (in final preparation). Persistent homology reveals meditation expertise-related regularity in auditory oddball EEG dynamics. Target journal: Neuroscience Letters Draft

Conference Papers & Proceedings

  • Rath, T., & Marupudi, V. (2025). Re-evaluating the numerical-perceptual distinction in the attraction effect. Proceedings of the 2025 Annual Meeting of the Cognitive Science Society (CogSci 2025). Link
  • Rath, T., Srivastava, N., & Srinivasan, N. (2026). A self-directed expanded judgment paradigm: Isolating the pairwise mechanism of the attraction effect. Accepted for oral presentation and full paper publication at the 2026 Annual Meeting of the Cognitive Science Society (CogSci 2026; 140/1511 oral presentations, 9.3%). Draft

Talks & Presentations

  • Rath, T., Srivastava, N., & Srinivasan, N. (2024, June). De-obfuscating context effects in decision-making: A simulation study. Paper presented at Virtual MathPsych/ICCM 2024. Link
  • Rath, T., Srivastava, N., & Srinivasan, N. (2024, December). Attribute trade-off difficulty modulates the asymmetric-dominance of the decoy. Presented at the Annual Conference of Cognitive Science, IIT Bombay.
  • Rath, T., Srivastava, N., & Srinivasan, N. (2024, November 22-25). Perceptual stimuli with difficult-to-trade-off attribute values show a positive attraction effect. Poster presented at the Annual Meeting of the Society for Judgment and Decision Making, New York City, USA.
  • Rath, T.. (in press). Effort-based strategy selection in multi-attribute decision making: When bounded rationality predicts optimal behavior. Applied Behavioral Science and Decision Making Conference (ABSDM 2025), IIT Mandi. PDF
  • Rath, T. (2026). Deliberation eliminates decoys rather than amplifying attraction. Poster to be presented at the 67th Annual Meeting of the Psychonomic Society, San Diego.

Awards & Research Support

  • Student Travel Grant, Cognitive Science Society (CogSci 2026) $1,500 USD
    Associated paper: A self-directed expanded judgment paradigm: Isolating the pairwise mechanism of the attraction effect. Awarded to a first author on a highly rated paper accepted for oral presentation.
  • Psychonomic Society Student Travel Award from Developing Nations $2,000 USD
    Associated abstract: Deliberation eliminates decoys rather than amplifying attraction. Selected for recognition at the 67th Annual Meeting of the Psychonomic Society.

CV

A compact CV is available here.

Selected Reads

A small working list of articles, repositories, and tutorials I have found useful for AI, cognitive science, and human decision-making.

Contact

I am open to postdoctoral research conversations in decision science, computational cognitive modeling, AI alignment, and human-AI interaction.

Email: tapasr@iitk.ac.in