Interaction between prior knowledge and type of nonlinear relationship on function learning
Abstract
The purpose of this thesis was to examine subjects' prediction patterns during training and during interpolation and extrapolation when subjects learned a function with different degrees of correspondence to prior knowledge. Experiments 1a and 1b showed that linear functions were easier to learn than power and other monotonic nonlinear functions. Experiment 2 indicated that prior knowledge interacted with function form during training. Experiment 3 replicated Experiment 2 but with an extrapolation test. One striking finding was subjects' tendency to fall back on prior knowledge when they made extrapolations even after they learned the training function very well. Finally, Experiment 4 examined whether or not subjects test hypotheses and found evidence for abrupt changes in hypotheses from trial to trial. None of the current major models accounts for the effects of prior knowledge on extrapolation adequately. An extension of the rule competition model (Busemeyer & Myung, 1992) seems to have potential to explain these results.
Degree
Ph.D.
Advisors
Busemeyer, Purdue University.
Subject Area
Psychology|Experiments|Cognitive therapy
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.