Surprise! Infants consider possible bases of generalization for a single input example

Abstract

Infants have been shown to generalize from a small number of input examples. However, existing studies allow two possible means of generalization. One is via a process of noting similarities shared by several examples. Alternatively, generalization may reflect an implicit desire to explain the input. The latter view suggests that generalization might occur when even a single input example is surprising, given the learner’s current model of the domain. To test the possibility that infants are able to generalize based on a single example, we familiarized 9-month-olds with a single three-syllable input example that contained either one surprising feature (syllable repetition, Exp. 1) or two features (repetition and a rare syllable, Exp. 2). In both experiments, infants generalized only to new strings that maintained all of the surprising features from familiarization. This research suggests that surprise can promote very rapid generalization.

The Distribution of Talker Variability Impacts Infants’ Word Learning

“…One possible explanation concerns the task complexity. Since infants at this age are willing to consider non-phonological acoustic dimensions as potentially relevant to word learning, they may have detected that talker gender was correlated with the words, increasing the complexity of the learning task (Gerken et al, 2015). On this view, infants needed to store two types of acoustic information with each word -talker gender and VOT.…”

“…On this view, infants needed to store two types of acoustic information with each word -talker gender and VOT. Research by Gerken et al (2015) suggests that storing two stimulus dimensions is more demanding on infant pattern learning than storing a single dimension. Task demands are also known to impact word learning (Fennell & Werker, 2003;Yoshida et al, 2009).…”

“…Structure on a non-phonological dimension that is usually not relevant to word learning-talker gender-appears to have increased the task complexity (Yoshida et al, 2009;Fennell, 2012;Werker & Curtin, 2005), perhaps by introducing competition between cues or creating a surprising correlation that infants needed to explain (Gerken et al, 2015). However, two other explanations are also possible.…”

Can infants learn phonology in the lab? A meta-analytic answer

“…But infants could draw spurious generalizations from even smaller training sets. In fact, recent work suggests that infants exhibit a significant preference between two words depending on whether they have recently heard a single exemplar that has low phonotactic frequency in the ambient language -showing that infants can generalize from a single type (Gerken, Dawson, Chatila, & Tenenbaum, 2015). In addition to making interpretation of previous work difficult, these results suggest that it may be impossible to investigate a general phonotactic pattern, for instance one that bears on a class of sounds, as the effective training set -regardless of how long the experimenter makes it -might be the last item (or that with the highest pitch, or the largest pitch range, or the highest breathiness...) heard by the child.…”

Learning multiple rules simultaneously: Affixes are more salient than reduplications

“…Second, infants expect items to conform to all generalizations they have picked up (see Gerken, Dawson, Chatila, & Tenenbaum, 2015, for an empirical confirmation of this point). As a result, they might consider triplets as a violation if any of the rules is violated.…”