S3E10 - Computational Rationality, Learning, Inference, & Bounded Bayes

S3E12 - The Future Boundaries of Rationality

Welcome to The Deep Dive: in this episode we unpack a modern paradox — astonishing scientific breakthroughs like DeepMind’s AlphaFold sitting alongside alarming declines in collective discourse and rising partisan polarization. We trace historical trends (the Flynn effect and its recent plateau), explore improvements in statistical and critical literacy, and weigh the dual role of AI as both a cognitive prosthetic that can amplify human reasoning and a ‘dark twin’ that can manipulate behavior. Featuring commentary from leading thinkers and researchers — including Steven Pinker, Nick Bostrom, John Kleinberg, Carl Friston, Philip Tetlock and others — the conversation surveys biological, computational and institutional limits to intelligence, debates superintelligence risks and benefits, and examines real-world examples from medicine to social media. Key themes include complementarity between humans and AI, algorithmic nudging and misalignment, the ‘dark forest’ of mistrust, and the practical levers for improving collective rationality. The episode concludes with practical foresight: three plausible future scenarios (a Rational Renaissance, muddling-through, or Collapse of Reason) and a seven-step high-stakes decision checklist designed to help listeners frame problems, gather evidence, check biases, apply tools, consider system effects, plan scenarios and implement adaptive feedback — a compact toolkit for making better individual and collective choices in an AI‑inflected century.

S3E11 - Debiasing & Rational Self-Regulation

In this Deep Dive episode the hosts explore why even highly intelligent people make predictable errors, unpacking major cognitive biases (anchoring, availability, confirmation bias, overconfidence, representativeness, framing, illusory correlation, authority bias) and the bias blind spot. They review evidence-based debiasing strategies — motivational, cognitive, and environmental — contrasting external nudges with self-nudging and discussing steering versus boosting. Practical tools covered include premortems, consider-the-opposite, the outside view, MCII (mental contrasting + implementation intentions), checklists, habit redesign, and feedback journaling. No external guests are featured; the hosts synthesize behavioral science research, real-world examples, and key takeaways about limits of debiasing, successful interventions, and building a durable personal toolkit for rational self-regulation.

S3E10 - Computational Rationality, Learning, Inference, & Bounded Bayes

In this episode, hosts unpack computational rationality by contrasting model-free and model-based learning through vivid examples — from a crashing delivery drone to Thorndike’s cats and Tolman’s cognitive maps. They explain how habits form as amortized inference, why planning saves costly trial-and-error, and how active inference reframes goals as prediction-driven behavior. The conversation covers reinforcement learning basics, the tradeoffs between fast habits and flexible planning, Friston’s free energy perspective, and the role of bounded optimality and policy complexity in shaping human decision-making. Practical takeaways include a "policy audit" for personal habits and a coding exercise to explore model-based vs. model-free agents.

S3E9 - Metacognition & Error Monitoring

In this episode of Deep Dive, the hosts unpack what happens in the mind when confidence outpaces reality, opening with Alex’s story — a trader whose rock‑solid gut led to a catastrophic loss. They trace the neural mechanics (ACC, ERN, DLPFC), explain how error monitoring and confidence judgments overlap, and discuss how internal consistency, prior beliefs, and the bias blind spot drive miscalibration. The conversation brings in insights from cognitive scientists and recent studies — from a UK survey on overconfidence to fMRI decoding of probabilistic brain states — and compares computational frameworks (signal‑detection versus Bayesian models). The episode also examines individual differences (age, genetics, expertise) and real‑world consequences across finance, medicine, and leadership. Finally, the hosts offer concrete, actionable strategies listeners can use immediately: run a personal prediction market game and score predictions with a Brier rule to train calibration, adopt a Bayesian mindset to treat beliefs as evolving probabilities, actively seek external critique to combat the bias blind spot, and use mindfulness and feedback to improve metacognitive accuracy. The takeaway: confidence can be measured, understood, and trained so your certainty better matches reality.

S3E8 - Anxiety, DMN, Hyperactivity, & Over-Simulation

In this episode of The Deep Dive, the host unpacks why high-functioning, analytical minds—exemplified by "Alex," a successful CEO—can become trapped in relentless worry loops. We explore the neuroscience of the default mode network (DMN), how predictive processing (priors and precision) and interoception drive rumination and physical anxiety, and how network dysfunction between the DMN, salience, and executive control systems produces decision paralysis. The episode cites leading research (Paulus & Stein, Buckner, Friston, Garrido and others) and surveys clinical findings in GAD and depression, then translates the science into practical interventions: SSRIs and beta blockers, CBT and mindfulness/ACT, exposure and biofeedback, and emerging approaches like psychedelic-assisted therapy. Key therapeutic targets and mechanisms are explained. Finally, the host offers an evidence-based, step-by-step DMN cool-down exercise (recognize and pause, belly breathing, 5-4-3-2-1 sensory scan, reality narration, re-engage, and optional worry appointment) so listeners can immediately interrupt over-simulation and regain present-moment control.