Quantum Cognition — Part 1 of “Quantum Approaches for Decision Making”
Program-level details: See program/curriculum.md
Combined 2026-06-01: This course is Part 1 of the combined 8-week, 4-credit course “Quantum Approaches for Decision Making” (Part 2 = Quantum Computing, Abhijeet Ghoshal). Decision: the two files remain as separate part-files (confirmed 2026-08-04 by Maria Rodas — see DECISIONS.md “Quantum course files: keep as two separate part-files”).
| Credits: 4 (combined course; this part ≈ Weeks 1-4) | Term: Spring 2027 (Weeks 1-4, from Jan 19 / POT A) | Instructor: Nathan Yang (Marketing, Associate Professor and C.W. Park Faculty Fellow) |
Website/Catalog Description (Combined Course — BADM 590)
Confirmed 2026-06-10 by Maria Rodas. This is the program-level description for the full 8-week course; Part 2 detail is in quantum_optimization.md.
This course pairs two quantum frameworks for working alongside AI, each running four weeks. In the first half, you use the mathematics of quantum probability — no physics or quantum hardware required — to model why human judgment systematically departs from the classical rationality of AI systems, learning to diagnose order effects, framing effects, and conjunction fallacies, then applying an audit methodology to AI-assisted decisions in your own organization. In the second half, you learn quantum computing from the ground up — core concepts, the underlying math, and the algorithms that solve complex business problems like optimization and route planning — building and testing hands-on solutions every week. AI tools accelerate exploration throughout. Portfolio artifact: An executive-ready decision audit and a quantum optimization solution.
Course Overview
Human judgment is not broken — it is structured differently than classical models assume. Quantum cognition applies the mathematical structure of quantum probability theory (no physics required) to explain systematic, predictable departures from classical rationality: preference reversals, conjunction fallacies, order effects, ambiguity aversion. For MSBAi students building AI systems, the stakes are direct: AI models are classically rational, but the humans who use and manage them are not. That cognitive asymmetry is where AI deployments fail in practice.
This course is Part 1 of Quantum Approaches for Decision Making (Spring 2027, Weeks 1-4), running four weeks with two 90-minute sessions per week. Part 2 is Quantum Computing and Decision Making (Weeks 5-8, Abhijeet Ghoshal).
Course Structure
| Week | Live Session (90 min) | Project Studio (90 min) |
|---|---|---|
| 1 | Module 1 — Why classical models break down. Detection versus construction; the measurement story and its hidden assumption; what a catalog of biases cannot do. | Company selection and problem framing. Instructor sign-off on scope. |
| 2 | Module 2 — Contextuality and order effects. Non-commuting judgments; the QQ equality as the load-bearing check that separates incompatibility from carryover. | Battery design and proposal workshop. Part 1 due end of week. |
| 3 | Module 3 — Human-AI decision teaming. Elicitation by systems; synthetic respondents; the audit statistics and the Tier 2 boundary. | Panel generation, audit troubleshooting, and peer critique of preliminary results. |
| 4 | Individual presentations and oral defense. Ten minutes each, followed by five minutes of questions. | Debrief and synthesis: what patterns held across audits, and where the framework did not apply. Part 2 due end of week. |
Live Session = Meeting 1 (Weeks 1-3, faculty content). Project Studio = Meeting 2 (Weeks 1-4, project work). Both Week 4 sessions devoted to project culmination. Live session attendance is not required or graded; Canvas engagement (one post + one peer response per week) is how engagement is verified.
Learning Outcomes (L-C-E Tagged)
| Learning Outcome | L-C-E |
|---|---|
| Explain, without mathematics, why quantum probability models human judgment better than classical probability in specific settings, and state what follows for a manager. | Literacy |
| Identify order effects, constructed preference, and interference in real organizational settings, and distinguish genuine non-classical structure from decorative invocation. | Literacy |
| Evaluate a human-AI decision process for cognitive asymmetry, using the framework as a diagnostic rather than a label. | Competency |
| Document how an analytical judgment changed across the project, including what prompted the revision and what the change revealed. | Competency |
| Design and execute a non-classical audit of an AI-generated synthetic respondent panel for a real firm and defend the findings under questioning. | Expertise |
Updated 2026-08-06 from Nathan Yang’s finalized syllabus (adapted from v12 proposal).
Assessment
| Component | Weight | AIAS | L-C-E | Description |
|---|---|---|---|---|
| Weekly diagnostic posts + Canvas engagement (×3) | 30% | 2 | L | One post and one peer response each week, verified through Canvas. Live session attendance not required or graded. |
| Project Part 1 — research proposal | 25% | 3 | C | Due end of Week 2. Identify a real company, state the management decision problem, design a three-item non-classical battery, and argue why order effects are expected. |
| Project Part 2 — audit report, deck, and oral defense | 35% | 4 (defense: 0) | E | Due end of Week 4. Field two synthetic respondent panels (classical + constructed), compute three audit statistics with bootstrap intervals, deliver a verdict and recommendation. 10-min individual oral defense + 5-min Q&A. AI is the specimen (AIAS 4); oral defense is AIAS 0. |
| Reflection memo (800 words) | 10% | 1 | E | After presentations: how analytical judgment changed across the project, what prompted revision, what the change revealed. |
All components are individual. Canvas engagement bundled into the 30% diagnostic posts; no separate participation line.
AI Attribution Log required on all project submissions. Graded component. Records tools, model versions, prompts submitted, and how judgment modified or overrode AI output.
Student effort: ~10-11 hrs/week (within MSBAi 8-12 hr target).
Updated 2026-08-06 from Nathan Yang’s finalized syllabus. Supersedes prior assessment structure (25% diagnostic / 25% milestones / 30% deck+defense / 10% reflection / 10% participation) drawn from v11/v12 proposal.
Content Modules (Live Sessions, Weeks 1-3)
Module 1 — Why Classical Models Break Down (Week 1) Detection versus construction; the measurement story and its hidden assumption; what a catalog of biases cannot do.
Module 2 — Contextuality and Order Effects (Week 2) Non-commuting judgments; the QQ equality as the load-bearing check that separates incompatibility from carryover.
Module 3 — Human-AI Decision Teaming (Week 3) Elicitation by systems; synthetic respondents; the audit statistics and the Tier 2 boundary.
Individual Project: A Synthetic Panel, Audited
The project runs in two parts, individual throughout. Students decide whether an AI-generated synthetic respondent panel can stand in for human respondents on a question a real company faces.
Part 1 — Research proposal (due end of Week 2): Identify a real, publicly documented company and state the management decision problem. Translate the problem into a research question. Explain why a synthetic panel is a candidate (budget, speed, or pretest). Argue from institutional detail why two specific framings plausibly compete — so that question order would move the answer. Name the strongest classical explanation and explain why it is not sufficient. Specify a three-item battery. State what the company should do if the audit comes back flat. Own employer permitted if supporting evidence is public.
Part 2 — Audit report, deck, and oral defense (due end of Week 4): Field two panels built from the same demographic grid — a classical panel (each item in isolation) and a constructed panel (all items in sequence with randomized order across respondents). Compute three audit statistics with bootstrap intervals: the order effect on the focal item, the QQ statistic, and the interference term. Deliver a verdict on whether the panel earns the right to stand in for people, and a recommendation the client could act on. Report where the Part 1 prediction held and where it did not.
Materials and Tools
Course materials: Selected chapters from Quantum Cognition for Management (Yang, manuscript), distributed through Canvas. No textbook purchase required. Also: underlying journal articles and case materials.
Tools: Python notebooks, VS Code or Colab, NumPy, access to a language model API. Assignments submitted as .ipynb files in a GitHub repository. No prior experience with quantum topics assumed.
Approval history: v12 approved by Vishal (2026-04-27); Maria Rodas coordinating submission to Amanda Brantner for formal academic approval. Finalized syllabus (adapted from v12) submitted by Nathan Yang 2026-08-06. Full decision record in the internal registry — available from K-ai; page history under “what changed?” above.