Cognitive Biases Yan Bulgak Mar 550 December 04
What Are They • A Cognitive Bias is the result of observation by a given person or persons that creates a skewed perception of “reality”
Confirmation Bias • Tendency to accept evidence confirming a held opinion while rejecting contradicting evidence • Underlies the fallacy of seeking scientific proof while not seeking disproval of alternative hypotheses • “You see what you expect to see” • The Murano™ Drivers Murano™ is a Registered Trademark of Nissan Corporation. Rules and restrictions apply. No substitutes. Only one per customer. Pre-approved credit available to select customer check store for details.
Selection Bias • Non-random selection of samples for experiment, thereby biasing the results • Rejection of “bad” data • Early trial termination • Medical trails – excluded data for patients who died or left. • See above Murano Drivers example • Evans proposed the ``Positivity Bias’’ – people confirm because they can’t think of how to falsify
Need For Closure… • …and the 2 am drunken phone call to the girl/boy friend you’ve not seen in 3 years. • People tend to avoid ambiguity on certain matters
Hyperbolic Discounting • Tendency to perceive immediate lesser gain as being more attractive than longer-term greater gain (even if greater gain is actually better for them)
Outcome Bias • …or the “I told you so” effect • Judging of past decisions on outcomes, though at decision time outcomes were uncertain • “Good researcher” is one who selected a topic that “worked”
Planning Fallacy • …or Cheops’s Law “Nothing is ever built on time or within budget” • Tendency to underestimate completion times
Self-Fulfilling Prophecy • Situation which initially false, becomes true through actions on the part of the actor. • Expectations of failure leads to failure
Notational Bias • A bias toward a particular range of expression that is circumscribed by the notation available • Western music: based on 12 tone equal temperament (TET) – ratio of • Ancient Chinese music used 7 (TET) • Arabic music uses 24 TET
Notational Bias (cont) • In Computer Science, a language that offers easy access to certain structures can define a pre-disposition toward a particular solution method that will carry over even into other languages • Examples: • PERL • Lisp and functional languages
Causes of Cognitive Bias • Attribution • Cognitive Dissonance • Heuristics (recall Troy’s presentation) • Representativeness • Availability • Anchoring
Attribution • Relies on Internal/External distinction in attributing blame/responsibility • Positivity Effect • Observer will attribute positive actions of a person they dislike to External factors, and negative actions to their internal disposition • Inverse for Negativity Effect • In general, people underestimate External effects
Attribution (Cont) • Other examples: • “Everything is easy to the man who doesn’t have to do it himself” • Backseat drivers • The “I told you so” effect
Cognitive Dissonance • Simultaneous expression of two contradictory beliefs • E.g. “I believe in the US Constitution” and “I am against free speech” • Holding such beliefs creates tension, which is reduced by changing the attitude toward one of the beliefs – the one that is least resistant to change.
Examples • George Orwell’s 1984 and “Doublethink” • MiniPax, MiniLuv, MiniTruth and MiniPlenty • Post-Purchase Rationalization
Effort Justification • Experimental evidence shows that subjects will rate initially undesirable objects higher if forced to suffer adversity to attain them. • Dissonance between: • Avoiding unpleasant effort • Mediocre outcome • Takes least effort to adjust evaluation of outcome. • Job Interviews • Military Bootcamp
Logical Fallacies • Post-hoc ergo propter hoc • Mean global temperatures are higher today than in 1800. The number of pirates today is smaller than in 1800. Therefore pirates and global temperatures are inversely correlated • Gambling expectation • Next ball will surely be red… • Murphy’s Law
The Texas “Sharpshooter” • Sharpshooter Story • Logical fallacy of adjusting hypothesis to outcome • Disconfirmation as strengthening cultist belief • Shakespearean Sonnets • and…
The Torah/Bible Code • Creates a space of letters by choosing every nthletter of the text • Look for “patterns” • Claim of future prophecies in the Pentateuch of the Torah • Rabin Assassination • 32 Famous Rabbis Experiment
Example • LoremipsumdolorsitametconsectetueradipiscingelitPhasellusunelitSedligulalacusportaveldignissimeususcipitsedfelisPhasellusetnibhutaugueconsequatvehiculaCrasfelissapiensemperafacilisissollicitudincondimentumeuismodorciNamauctorlobortiselitCraseliterosrhoncusacauctorvit • N = 5 • LoremipsumdolorsitametconsectetueradipiscingelitPhasellusunelitSedligulalacusportaveldig • LidsestdclauldlutdsuieissbuouhafsneciluoeeoiultisehsciasaainifietsenauvIminuutn
Concerns and Refutations • On the non-statistical side: • Clear example of Confirmation and Selection bias • Statistical side: • McKay et al showed that Rabbi experiment highly sensitive to minor changes • Liberties and choice of Rabbi names • Other linguistic features, when altered, reduce statistical likelihoods by many orders of magnitude • Replicated experiment with War And Piece can be made to yield significance results of 1 in a million!
Bibliography • Brock, T. C. and M.C. Green (Eds). Persuasion: Psychological Insights And Perspectives 2nd Ed. Thousand Oakes, CA: SAGE Publications, 2005 • Drosnin, M. The Bible Code. New York, NY: Simon & Shuster, 1997 • Evans, Jonathan St. B.T. (Ed). Bias In Human Reasoning: Causes And Consequences. Hove, UK: Lawrence Erlbaum Assoc, 1989 • Kahneman, D., Slovic, P. and A. Tversky (Eds). Judgment under Uncertainty: Heuristics and Biases. Cambridge, UK: Cambridge University Press, 1982. • McKay B., Bar-Natan D., Bar-Hillel M., and G. Kalai. Solving The Bible Code Puzzle. Statistical Science Vol.14 150-170, 1999 • http://en.wikipedia.org/wiki/List_of_cognitive_biases • WitztumD, Rips E. and Y. Rosenberg . Equidistant Letter Sequences in the Book of Genesis. Statistical Science Vol.9 429-438, 1994