Decision Sciences: Cliff's Notes Refresher
A practitioner's overview of the academic field, anchored in the Georgia State University tradition, with deep-dive prompts for any topic you want to revisit.
---
How to use this document
Each topic has two kinds of links:
- 📖 Learn more — links to authoritative external sources (Wikipedia, Stanford Encyclopedia of Philosophy, peer-reviewed articles, primary texts).
- 🤖 Go deeper with Claude — one-click prompts that open a fresh Claude conversation set up to generate an in-depth tutorial on that specific topic. Treat these as your "expand this section" buttons.
---
1. What is Decision Sciences?
Decision Sciences is the interdisciplinary study of how decisions are (and should be) made under uncertainty, complexity, and limited cognitive capacity. It sits at the intersection of:
- Operations Research / Management Science — quantitative optimization, simulation, queuing, linear programming
- Behavioral Economics & Cognitive Psychology — how humans actually decide vs. how rational-choice theory says they should
- Statistics & Data Analytics — inference, forecasting, Bayesian reasoning
- Information Systems — decision support systems (DSS), expert systems, now AI/ML
- Philosophy of Action — utility theory, ethics, the very definition of "a decision"
The field's organizing question, articulated most clearly by Herbert Simon: "How do agents with bounded cognitive resources make tolerably good decisions in environments more complex than they can fully understand?"
- 📖 Decision Sciences Institute · Wikipedia: Decision theory
- 🤖 Generate a 2,000-word primer on the field
---
2. The Georgia State University Context
Your concentration sat inside what was historically the Department of Decision Sciences at the J. Mack Robinson College of Business. Curriculum-wise, Robinson's MBA decision-track courses traditionally drew from:
- Quantitative methods in business (statistics, regression, optimization)
- Decision theory (normative & descriptive)
- Decision support systems and group decision support
- Individual and group problem solving
- Utility theory
In the years since, GSU has reorganized — the Decision Sciences department has been merged into the Department of Managerial Sciences, and decision-oriented coursework now also appears in Robinson's MS in Business Analytics, the STEM-designated MBA in Business Analysis, and the AI Business Innovation graduate certificate. The intellectual lineage is the same; the labels have shifted toward "analytics" and "data-driven decision-making."
The MBA core course MBA 8125 / "Data Driven Decision Making" is the modern descendant of the foundation you took.
- 📖 Robinson College of Business — MBA · Robinson Departments
- 🤖 Map your old GSU coursework to today's curriculum
---
3. The Theoretical Bedrock
3.1 Bounded Rationality (Herbert Simon, 1955)
Simon's central insight: humans are not the omniscient utility-maximizers of classical economics. We have limited information, limited computational capacity, and limited time. So instead of optimizing, we satisfice — search until we find an option that meets a threshold of "good enough," then stop. Simon won the Nobel for this in 1978 and the Turing Award (with Newell) in 1975.
Why it matters: every modern decision tool is in some way a workaround for bounded rationality — either compensating for it (decision support systems) or exploiting it (nudges, choice architecture).
- 📖 Stanford Encyclopedia: Bounded Rationality · Wikipedia: Bounded rationality
- 🤖 Deep dive into bounded rationality
3.2 Simon's Decision Process Model (your "3-phase model")
Originally three phases (1960), Simon later added a fourth:
| Phase | What happens | Modern toolkit |
|---|---|---|
| Intelligence | Scan environment, recognize a problem/opportunity exists, gather data | Dashboards, KPI alerts, environmental scanning, SWOT, PESTEL |
| Design | Generate alternative courses of action, model their consequences | Brainstorming, scenario planning, simulation, decision trees |
| Choice | Evaluate alternatives against criteria, select one | Decision matrix, cost-benefit, expected utility, sensitivity analysis |
| Implementation | Execute, monitor, feed results back into intelligence | Project management, KPI tracking, post-mortems |
The model is iterative — you loop back constantly. It's still the spine of most rational decision frameworks taught today.
3.3 Heuristics & Biases (Kahneman & Tversky)
Where Simon said "humans satisfice," Kahneman and Tversky catalogued exactly how — and where it goes wrong. Key ideas:
- System 1 / System 2 — fast, automatic, intuitive vs. slow, effortful, deliberate
- Prospect Theory — losses loom roughly 2x larger than equivalent gains; we're risk-averse for gains, risk-seeking for losses
- Anchoring, availability, representativeness, framing — the canonical biases
Every decision-maker should know the top dozen biases the way a doctor knows the top dozen drug interactions.
---
4. Decision Frameworks (the "which model when" toolkit)
4.1 The Rational / Normative Model
Define problem → identify criteria → weight criteria → generate alternatives → score → choose. Best for structured, low-time-pressure decisions with quantifiable trade-offs.
4.2 OODA Loop (Observe → Orient → Decide → Act)
Boyd's military framework. Best for high-velocity, adversarial, ambiguous environments. The advantage goes to whoever cycles the loop faster — "getting inside the opponent's OODA."
4.3 Cynefin Framework (Snowden)
A meta-framework: classify the situation first, then pick your tool.
- Clear/Obvious → sense, categorize, respond (best practice)
- Complicated → sense, analyze, respond (good practice, expert-driven)
- Complex → probe, sense, respond (emergent practice, experiments)
- Chaotic → act, sense, respond (novel practice, stop the bleeding first)
- Confused/Disorder → break it apart into the above
This is the single most useful "which framework do I even use" tool ever created.
- 📖 Wikipedia: Cynefin framework · HBR: A Leader's Framework for Decision Making
- 🤖 Diagnose your situation with Cynefin
4.4 Vroom-Yetton-Jago
For leadership decisions: should I decide alone, consult, or decide with the group? A decision tree based on decision quality, acceptance needed, and time pressure.
4.5 RAPID / DACI
For organizational decision rights. Who Recommends, Agrees, Performs, has Input, Decides? (Or the simpler DACI: Driver, Approver, Contributor, Informed.) Solves the "who actually owns this?" problem in big orgs.
---
5. Analytical Tools (the math)
| Tool | Use it when… | Output |
|---|---|---|
| Decision Tree | Decisions have sequential branches with probabilities | Expected value of each path |
| Weighted Decision Matrix | Comparing 3+ options across multiple criteria | Ranked, scored options |
| Cost-Benefit Analysis | Trade-off is reducible to dollars (or a single utility) | NPV, ROI, payback |
| Expected Utility | Outcomes are probabilistic and you have a utility function | Best-bet option |
| Sensitivity Analysis | You want to know which assumptions matter most | Tornado chart |
| Monte Carlo Simulation | Many uncertain inputs interact | Distribution of outcomes |
| Linear/Integer Programming | Constrained optimization (resources, scheduling) | Optimal allocation |
| Real Options Analysis | Decisions are staged and reversible | Option value of waiting |
| Bayesian Updating | You'll learn more as the decision unfolds | Posterior probabilities |
---
6. Diagnostic / Problem-Definition Tools
These help in Simon's Intelligence phase — making sure you're solving the right problem.
- Fishbone / Ishikawa Diagram — root-cause analysis across 6Ms (Methods, Machines, Materials, Manpower, Measurement, Mother Nature) or 4Ps (People, Process, Policy, Plant). Born in Japanese quality control, still indispensable.
- 5 Whys — Toyota's radically simple root-cause technique. Ask "why?" five times in succession.
- Pareto Analysis (80/20) — find the vital few causes producing most of the effect.
- SWOT / PESTEL / Porter's Five Forces — environmental scanning at strategic level.
- Issue Tree / MECE decomposition — McKinsey's bread and butter; break a problem into Mutually Exclusive, Collectively Exhaustive sub-problems.
- Jobs-to-be-Done (Christensen) — reframe the problem from the customer's progress, not your product's features.
---
7. Group & Organizational Decision Tools
- Six Thinking Hats (de Bono) — assigned-perspective brainstorming: White (facts), Red (feelings), Black (risks), Yellow (benefits), Green (creativity), Blue (process).
- Delphi Method — anonymous, iterative expert forecasting; converges without groupthink.
- Nominal Group Technique — silent generation → round-robin → discussion → vote. Beats free brainstorming on idea volume and quality.
- Pre-Mortem (Klein) — "Imagine it's a year from now and this decision was a disaster. What happened?" Devastatingly effective at surfacing risks groupthink suppressed.
- Red Team / Devil's Advocate — formal dissent role.
- Decision Journals — write down your prediction and reasoning before outcomes land, to combat hindsight bias.
---
8. Where the Field Is Headed: Decision Intelligence
The current research and practice frontier blends classical decision sciences with modern AI/ML:
- Decision Intelligence (DI) — Cassie Kozyrkov (formerly Google) coined the umbrella term. Treats decisions as engineering artifacts: modeled, instrumented, A/B-tested, and improved.
- Causal Inference — Pearl's do-calculus, DAGs, counterfactual reasoning. Moves analytics from "what correlates" to "what would happen if."
- Reinforcement Learning for Sequential Decisions — Markov Decision Processes, policy optimization.
- AI-Augmented Decision Support — LLMs as Socratic partners for the Intelligence and Design phases (this is exactly what the companion tool below is built for).
- Algorithmic Auditing & Decision Ethics — fairness, accountability, transparency in automated decisions.
- 📖 Cassie Kozyrkov on Decision Intelligence · Judea Pearl — The Book of Why
- 🤖 Get oriented to Decision Intelligence
---
9. Recommended Reading (in order of accessibility)
- *Daniel Kahneman — Thinking, Fast and Slow*** — the gateway drug
- *Annie Duke — Thinking in Bets*** — decisions vs. outcomes; resulting bias
- *Chip & Dan Heath — Decided*** — the WRAP framework, very practical
- *Gary Klein — Sources of Power*** — naturalistic decision-making, pre-mortem
- *Herbert Simon — Administrative Behavior*** (1947) — the original
- *Howard Raiffa — Decision Analysis*** — the classical textbook
- *Judea Pearl — The Book of Why*** — causal revolution
- *Cass Sunstein & Richard Thaler — Nudge*** — choice architecture
---
10. The companion tool
The interactive React tool that ships with this overview is built around Simon's 4-phase model as its spine, with Cynefin doing the situation-classification up front and a library of the tools above pluggable into each phase. It uses the Claude API to act as a Socratic partner, not an oracle — it asks questions, surfaces biases, and helps you produce a written decision brief. The goal is not to make the decision for you. The goal is to make sure you've done the thinking.