March 16, 2010 Psychology 485 http://www.youtube.com/watch?v=tvaD- 29pQBY Introduction • History & Definitions Self-recognition • Associative processes? Metacognition • Associative processes, modeling and Behavioral Economics Rene Descartes • Cogito ergo sum • I think, therefore I am Dualism Cartesian theatre • A place in your head where “you” are watching things happen Humans are aware of ourselves as animate beings • Control of own behaviour • Mental representation of ourselves Are animals self-aware? • Mirror tasks Thinking about thinking Primary vs secondary representations Assessing internal states is not enough • Knowing that you are hungry isn’t metacognition Assessing knowledge states Know a lot • Some people know a lot about baseball, some don’t know much Do you know how much you know? John is a moderate fan of baseball • e.g. “I really have to study for this midterm tomorrow, I don’t know anything!” Know a little Do not interpret as higher cognitive process if lower process will suffice Difficult to “show” secondary representations (especially without language) • Can self-awareness and metacognition be explained through reinforcement history and/or associative learning? A test of self-recognition, self-consciousness Stages: • Time to adjust/experience mirror • Tranquilize animal and paint 2 dots (visible and controlhidden) • See if animal notices dot, compare to control dot Animals cats? tested: chimps, dolphins, elephants, magpies, Epstein, Lanza (1981) & Skinner • Trained pigeons to peck at blue dot • Experience with mirror (see blue dot in mirror, peck at origin) • Blue dot on pigeon, under bib • Peck at bib video Skinner Kinds of questions we ask children reinforces self-observation e.g., “are you hungry?” “what are you doing?” Accurate response likely results in some form of desired outcome (i.e., reinforcement of behaviour) 11 Do animals know when they don’t know? • Dolpins, pigeons, rats, non-human primates Testing procedure • Some trials include the option to ‘decline’ • If animals know they don’t know, should decline to answer Study phase: Short or Long tone 0.33 0.66 Choice phase: 1/3 Forced Test 2/3 Choice Test phase: 6 pellets if correct 0 pellets if incorrect 3 pellets If animals have metacognition: • Increase use of ‘decline’ option as task difficulty increases Red-green not much use of ‘decline’ Light green-dark green more use of ‘decline’ • Accuracy is higher on ‘chosen’ tests than ‘forced’ tests You choose to take the test when you know the answer • Accuracy difference increases with task difficulty Can associative processes explain higher accuracy on ‘Chosen’ tests? Smith, Beran, Couchman, & Coutinho, 2008 • Reinforcement of ‘decline’ options creates a “low frequency tendency” to decline • Competes with generalization gradients for each stimulus High Response Strength Decline Threshold Low Short Subjective level of stimulus Long Reinforcement to decline option creates a constant response-strength tendency • Competes with response-strength of stimuli Winner-take-all mechanism Since it is based on subjective view of stimuli, also accounts for difference between Chosen-Forced accuracy Shows associative processes can explain metacognition Morgan’s canon? Jozefowiez, Staddon & Cerutti, 2009 Similar to quantitative model, but measures • Probability of payoff • Risk levels (is animal risk-prone or risk-averse?) 1.0 Short response Long response Probability of payoff at subjective equality is diminished Payoff 0 Short Subjective level of stimulus Long 1.0 Short response Long response Correct 50% of time, average reward = 3 pellets Decline reward = 3 pellets Payoff Risk Neutral 0 Short Subjective level of stimulus Long 1.0 Short response Risk Averse Payoff Long response Would rather guarantee payoff of 3 than risk no reward 0 Short Subjective level of stimulus Long More on this next week... When might an animal want to guarantee some kind of payoff? When might they be willing to “risk it” for the larger payoff? Model accounts for changing needs, and metacognition • Still doesn’t assume metacognition http://www.zefrank.com/theshow/archiv es/2007/01/010807.html Is selfawareness/metacognition/consciousness necessary? • Why learn to be self-aware? Evolutionary advantages?