Do Rational People Exist

  • Is there a subpopulation of individuals who consistently exhibit less cognitive bias and do better with judgement under uncertainty than the median person?
  • No cognitive bias affects 100% of all people
  • Correlation between cognitive bias tasks
    • Cognitive reflection test correlates with the cognitive bias test, which also correlates with IQ
    • However, many other tests of cognitive bias fail to correlate with IQ
  • Cognitive bias performance and dispositions
    • Dispostions are personal qualities that reflect one’s priorities
    • Performance on cognitive bias tests is correlated with the sorts of dispositions that are associated with being a thoughtful and reasonable person
  • Factors correlating with cognitive bias
    • The group that got all three questions wrong on the cognitive bias test was 66% women
    • Professional scientists and undergraduates did no better than science and math undergrads, but undergrad science and math students outperformed history undergraduates
    • Autistics are less susceptible to the conjunction fallacy
    • Higher executive function appears to correlate with avoiding intuitive-but-wrong answers
    • The ability to switch tasks and avoid being anchored to pre-existing beliefs correlates with prefrontal cortex learning
  • Is there a hyper-rational elite?
    • People who are rational in one area tend to be more rational in other areas
    • Rationality appears to correlate with the things you’d expect it to correlate with
      • Intelligence
      • Mathematical ability
      • Flexible cognitive dispositon
    • Tetlock’s experiments suggest that there may exist a small population of outliers with extremely low rates of cognitive bias
    • Doesn’t yet appear to be evidence that superforecasters score exceptionally high on a variety of cognitive bias tests
  • Stanovich’s tripartite model
    • Stanovich proposes dividing System II into two parts
      • Reflective mind - self-skepticism; interrupts thinking and asks “is this right”
      • Algorithmic mind - working memory and cognitive processing capacity - measured by IQ
    • The reason that some cogintive biases but not others correlate with IQ is because some cognitive biases are handled by the algorithmic mind, whereas others are handled by the reflective mind
      • Intelligence doesn’t protect against sunk-cost fallacy, bias blind spots, myside bias or anchoring effects
      • Intelligence is correlated with various tests of probabilistic reasoning, logical reasoning, and maximization of expected value
    • Stanovich talks about cognitive decoupling and cognitive flexibility
      • Cognitive decoupling: ability to block out experiential knowledge and follow formal rules
      • Cognitive flexibility: ability to question your own beliefs
    • People with high IQ are good at cognitive decoupling, but not necessarily cognitive flexibility
  • Conclusions
    • There are people who are less cognitively biased than average
    • However, we can’t tell if there is a tiny minority of extremely rational people because our tests don’t have enough experimental power to detect the extremes of performance on tests of cognitive bias

Critical Thinking

  • Everyone agrees that more critical thinking skills are beneficial
  • However, can critical thinking really be taught?
    • Process of critical thinking is irrevocably intertwined with content of thought
    • Can’t look at an issue from multiple perspectives if you don’t have access to the other perspectives
  • Why is thinking critically so hard?
    • Appears to be extremely context sensitive - people who display excellent critical thinking in one situation display an utter lack of critical thinking in other scenarios
    • People get lost in surface structure differences and don’t notice deep structure similarities
  • In order to notice deep structure similarities, a person needs the following two prerequisites:
    • Knowledge that they should look for deep structure
      • Comes from metacognitive knowledge
      • While knowing that one must look for deep structure can be helpful, it is of limited utility, since it doesn’t tell you what the deep structure is
    • Familiarity with a problem’s deep structure
      • Familiarity comes from repeated exposure
      • Need to have handled different manifestations of the same problem
      • After repeated exposure to different variants of the same problem, students immediately see the deep structure when presented with a novel variant
  • Is thinking like a scientist easier?
    • Scientific reasoning is a subset of reasoning that is not all that much different from other types of reasoning
    • What makes it “scientific” is knowing that one must engage in it
    • Science is a constrained content area, which makes it easier to look for cues that tell you what metacognitive strategies to use
    • Greater background knowledge and experience with a field allows you to devise more effective experiments and better interpret their results
    • Scientific thinking must be taught alongside scientific content

Do Lifehacks Ever Reach Fixation

  • Lifehacks occupy a restricted range
    • Not completely useless, otherwise we wouldn’t have heard about them
    • Not completely useful, otherwise we’d already be doing them
  • Why haven’t lifehacks reached fixation?
    • Potential hypotheses
      • No exceptionally good lifehacks - the human body and brain are already pretty well optimized
      • Lifehacks, as a category, have some characterisitic that makes fixation an unreasonable goal - maybe people are diverse enough that no single life-hack will ever work for everyone
      • Lifehacks haven’t existed long enough to reach fixation
      • All of the genuinely useful lifehacks take too much work
      • Some life hacks have reached fixation, and we don’t see them because they’re omnipresent
    • The last hypothesis is the most interesting, so let’s try to think of some lifehacks that have reached fixation
    • Look for things that act directly to raise energy levels, intelligence, social skills, or organizational ability
  • Writing things down to remember them
    • To-do lists
    • Three-ring binders
  • Diet and exercise
  • Take a deep breath and count to 10
  • Bags (and other containers)
  • Caffeine

Superforceasting: How to Upgrade Your Company’s Judgement

  • The type of forecast that’s most amenable to improvement is forecasts for which some data, logic and analysis can be used, but where seasoned judgement and careful questioning are also important
  • How to improve forecasts
    • Train for good judgement by learning about reasoning errors and cognitive biases
    • Build the right kinds of teams
      • Composition
        • Focus on humility
        • Intellectual diversity - get a mix of experts and intelligent and motivated non-experts willing to challenge the experts
        • Diverging, evaluating, and converging
          • Diverging and evaluating phases are critical to ensure the team doesn’t tunnel-vision on a single forecast or scenario
          • Narrow ranges often reflect overconfidence
        • Team members need to trust each other
    • Track performance and give feedback
      • There is no way to improve predictions if you don’t keep track of how well the predictions corresponded with reality
      • When comparing a prediction to the result, also look at the process used to generate that prediction, and examine it for systemic biases