Evidential diversity increases generalisation in predictive learning

Abstract

In property induction tasks, encountering a diverse range of instances (e.g., hippos and hamsters) with a given property usually increases our willingness to generalise that property to a novel instance, relative to non-diverse evidence (e.g., hippos and rhinos). Although generalisation in property induction and predictive learning tasks share conceptual similarities, it is unknown whether this diversity principle applies to generalisation of a predictive association. We tested this hypothesis in two predictive learning experiments using differential training where one category of stimuli (e.g., fruits) predicted an outcome and another category (e.g., vegetables) predicted no outcome. We compared generalisation between a Non-Diverse group who were presented with non-diverse evidence in both positive (predicted the outcome) and negative (predicted no outcome) categories, and two groups who received the same training as the Non-Diverse group but with a more diverse range of exemplars in the positive (Diverse+ group) or negative (Diverse– group) category. Diversity effects were found for both positive and negative categories, in that learning about a diverse range of exemplars increased generalisation of a predictive association to novel exemplars from that same category. The results suggest that diversity, a key principle describing how we reason inductively, also applies to generalisation in associative learning tasks.

Individual differences in causal structures inferred during feature negative learning

“…The causal learning literature shows that participants interpret associations differently according to the causal model implied by the cover story (Waldmann et al, 2006; Waldmann & Holyoak, 1992), but there have been few attempts to investigate whether different causal models (structures) arise from associative learning procedures with more neutral cover stories. Such a connection would further strengthen the view that associative learning theories can account for how humans learn about causal relationships (Dickinson et al, 1984) and help clarify the role of higher-order cognitive processes in human associative learning (Lee, Lovibond, & Hayes, 2019; Lee, Lovibond, Hayes, & Navarro, 2019; McLaren et al, 2018; Mitchell, De Houwer, & Lovibond, 2009).…”

Stimulus discriminability and induction as independent components of generalization.

“…Cognitive accounts, by contrast, have emerged from human research showing that generalization can be guided not only by similarity but also by factors such as category knowledge (Dunsmoor, Martin, & LaBar, 2012; Dunsmoor & Murphy, 2014; Lee, Lovibond, & Hayes, 2019), symbolic relations (Boddez, Bennett, van Esch, & Beckers, 2017; Dymond, Schlund, Roche, & Whelan, 2014), and verbal instructions (Vervliet, Kindt, Vansteenwegen, & Hermans, 2010). Our own work has revealed qualitative differences in generalization gradients between subgroups of participants depending on their inferred rules (Lee, Hayes, & Lovibond, 2018; see also Wong & Lovibond, 2017).…”

Inhibitory summation as a form of generalization.

“…However, objects that share similar underlying properties often also share a degree of physical similarity (Lee et al, 2019; Rosch & Mervis, 1975). A potential way to manipulate stimulus similarity independent of physical similarity is to take advantage of the acquired equivalence phenomenon described earlier (Dymond et al, 2012; Honey & Hall, 1989; Hall et al, 1993).…”