Static Human Gaze Activity Does Not Diminish Infants’ Cognitive Load during Referential Learning

Poster Presentation: Sociolinguistic Variation and Language Processing Conference. Blacksburg, Virginia. Thursday 31st March, 2016.

Hans Carlo Rivera, Molly Kellington, Laura Mills-Smith & Robin Panneton. iLEAP Lab, Department of Psychology, College of Science, Virginia Polytechnic Institute and State University

Abstract

Werker et al. (1998) habituated 14-month-olds to two object-label pairings (ObjectA+LabelA and ObjectB+LabelB), and then tested infants’ discrimination of a subsequent mismatch (ObjectA+LabelB; called the ‘switch’). When the labels were maximal pairs (e.g., lif v. neem), infants significantly recovered attention on the switch trial. In contrast, Stager and Werker (1998) found no significant recovery of attention on switch trials with minimal pair labels (bih v. dih), suggesting that the cognitive resource requirement of the task is essential to referential learning. In fact, lessening the cognitive demand (e.g., familiarizing infants with minimal pairs in advance of the task; Fennel & Werker, 2003) augments infants’ ability to learn difficult (e.g., doll and ball) object/word pairings. The purpose of this study was to examine whether the inclusion of a female face, making eye contact with the object being labeled, would also act to decrease cognitive load in a minimal pair referential task.

Using an infant-controlled switch task, a total of 18 13- to 15-month-olds were presented with object-label pairings in one of two conditions: maximal pairs (e.g., jode and pake) or minimal pairs (e.g., bin and din). Importantly, a female face first appeared on the screen, looking directly at the infant, then shifted gaze to the right (or left) in advance of the object+label presentation (on the corresponding side of the gaze shift). Infants’ looking times were measured and compared across test trials and groups, resulting in a significant trial type x group interaction (F(1, 16)=7.12, p=.02, ηp2=0.31). Infants in the maximal pair group looked longer on switch trials (Mswitch=10.13, SD=1.8) than same trials (Msame=6.83, SD=1.03), whereas infants in the minimal pair group did not (Mswitch=4.76, SD=1.76; Msame=6.85, SD=1.46). Importantly, no significant difference in looking times on the objects and faces was found within each condition (Minimal, p=.20; Maximal, p=.37), nor between conditions (Faces, p=.83; Objects, p=.88).

These results replicate the general finding that as cognitive load increases (e.g., minimal pair object labels), infants have more difficulty learning object-label relations. The addition of a female gaze-relevant reference did not decrease this cognitive load, in spite of the value of gaze contingency for other aspects of early language learning (e.g., Senju, Csibra, & Johnson, 2008). It is possible that in order for human gaze-relevancy to attenuate cognitive load in referential tasks, other ostensive forms of communication must also be present (e.g., smiling; the label emanating directly from the agent). It also possible that mutual gaze was not established in the current study given no formal contingency between where infants’ looked before gaze shift was implemented. In this vein, Fais et al. (2012) demonstrated that infants’ success with minimal pairs in referential learning was positively associated with sharing mutual gaze with a live experimenter, suggesting that this may be related to emerging joint attention skills. We will discuss ways in which our own “social switch” protocol can be improved to directly test this hypothesis (e.g., the use of gaze-contingent eye tracking) as well as other ostensive cues.