Underestimation in linear function learning: Anchoring to zero or x-y similarity?

Function learning research has shown that people tend to underestimate positive linear functions when extrapolating Y for X-values below the training range. Kwantes and Neal (2006) proposed that this underestimation occurs because people anchor their Y-estimates at zero. It is equally plausible, however, that people are biased to make Y-estimates similar to the presented X-value. To differentiate these 2 explanations, 135 participants extrapolated positive linear functions with a y-intercept either greater than or less than zero. In line with the anchoring hypothesis, participants underestimated in the lower extrapolation region when the y-intercept was positive, but overestimated when the y-intercept was negative. These results are consistent with a version of the extrapolation association model (EXAM; Delosh, Busemeyer, & McDaniel, 1997), which proposes that people interpolate linearly between the training exemplars and zero in the lower extrapolation region. (PsycINFO Database Record