Prior knowledge and correlational structure in unsupervised learning.

Abstract

Prior knowledge has been shown to facilitate both supervised and unsupervised category learning, but questions remain about how this facilitation occurs. This article describes two experiments that investigate the effects of prior knowledge on unsupervised learning, using the exemplar-memory task of Clapper and Bower (2002). Experiment 1 demonstrates that prior knowledge facilitates learning in this task, as expected, and that this facilitation extends to both knowledge-relevant and knowledge-irrelevant features of the new categories. Experiment 2 shows that knowledge facilitates learning not only by increasing the probability that people will discover separate categories, but also by making the features of different categories seem less interchangeable, thereby reducing interference and confusion among them. Taken together, these experiments demonstrate that prior knowledge has multiple effects on unsupervised learning and suggests that the exemplar-memory task may provide a useful procedure for disentangling and investigating these effects.

The effects of prior knowledge on incidental category learning.

“…The present demonstration of a knowledge effect on neutral features provides evidence for such a category cuing effect and reinforces similar results reported by Kaplan and Murphy (1999) and Clapper (2007).…”

“…Another difference from Experiment 1 is that the difference between thematic and neutral dimensions failed to reach significance in the opposite-theme condition of the present experiment, i(10) = 0.50, despite a small difference in means in the expected direction. The reason for this non-result is unclear, particularly as significant differences between thematic and neutral feature were found in both Clapper (2007) and the present Experiment 2.…”

“…As noted in the introduction. Clapper (2007) reporied data suggesting that the feature facilitation effect in the exemplar-memory task is largely, if not entirely, a segregation effect rather than a binding effect. The experiment described next is another attempt to disentangle and measure these two potential sources of feature facilitation.…”

“…This improvement, relative to a control condition in which the training instances lack any consistent category stmcture, reveals the person's leaming. Clapper (2007) demonstrated that prior knowledge could facilitate unsupervised leaming in this task. In particular, when the correlated features that defined two categories were related to familiar "themes" (e.g., suggesting that members of one category were snobby "highbrows" and those of the other were down-toearth "lowbrows"), people showed much better leaming (memory for consistent features) than if the categories did not relate to contrasting themes or stereotypes.…”

La Edad Se Correlaciona Directamente con la Fuerza de los Estereotipos de Género: Evidencia Obtenida en una Tarea de Memoria de Reconocimiento

“…En algunos casos el concepto puede hacer el procesamiento más eficiente. Por ejemplo, durante el aprendizaje de un concepto nuevo, el hecho de poseer un concepto previo relacionado lo hace más fácil de aprender (Clapper, 2007;Hoffman, Harris & Murphy, 2008;Kaplan & Murphy, 2000;Murphy & Allopenna, 1994). Sin embargo, y al mismo tiempo, poseer un concepto puede introducir errores o sesgos (e.g., Elischberger, 2005).…”