Remembering Words During Sleep: Behavioral and Neural Indices

• In Event: 3-188 - Sleep, Memory and the Brain During Development

Integrative Statement

Introduction and hypotheses:
Episodic memory, the capacity to remember the past with specific detail, emerges in infancy and develops considerably during early childhood (Bauer, 1996). Early changes in hippocampal function have long been suggested to underlie these improvements, and our laboratory provided initial evidence connecting hippocampal function and memory in young children (Prabhakar et al., 2018). It has been also suggested that the hippocampus plays an integral part in the initial acquisition of novel vocabulary (Warren & Duff, 2014), a process requiring binding arbitrary words and their referent objects, similar to binding processes supporting episodic memory. Our current study seeks to examine whether the hippocampus supports initial word learning during a period of substantial vocabulary expansion.

Methods and Results
Participants included 36 24-32-month-old toddlers, who completed two separate word-learning tasks, and a functional MRI task during sleep.
In the object-label task, toddlers were briefly exposed to toys made from colorful foam, two of which were well known objects (e.g., cup) and two of which were unknown objects with novel shapes. Toddlers were taught the label of one of these novel objects, which was a pseudoword (e.g., “Croy”). Toddlers were then asked to hand over the “Croy” by selecting it from 4 objects including the target and 3 distracters, which were also in view during learning. Toddlers underwent this procedure twice to learn to words. Toddlers successfully learned the labels at above-chance levels (1/4) levels (M=.69; SD=.37), t(33) = 6.96, p < 0.001. After a 3 – 8 day interval, toddlers completed the same test without re-exposure and continued to show above-chance accuracy (M=.53; SD=.39), t(32) = 4.09, p < 0.001 (Fig. 1).
In the character-name task, children were exposed to pseudowords in a more ecologically-minded experience. Toddlers participated in a play session involving an animal puppet whose name was a pseudoword (e.g., “Grom”). An additional puppet “lived” in the room in plain sight but was not given a label. Toddlers were asked to hand “Grom” to the experimenter by selecting it from 4 puppets, including the target, the non-labeled puppet, and 2 additional novel animal puppets. Toddlers underwent this procedure twice and performed above chance (M=.48, SD=.42), t(30) =3.12, p < 0.01. After a 3-to 8-day interval without any additional exposure to the label. Toddlers successfully still remembered the label above-chance levels after the delay (M=.62, SD=.41), t(29) = 4.92, p < 0.001 (Fig. 1).
On the night of the last session, a subset of these 2-year-olds (N= 13; projected N=24) participated in an fMRI task, in which they heard 3 20-second blocks (learned object pseudowords, learned character pseudowords, novel pseudowords) during natural nocturnal sleep. We found a significant cluster of activity for learned labels for objects > unknown pseudowords in the anterior hippocampus (Fig. 2). Activation for this contrast was correlated with performance both on the object task r(12) = 0.601, p = 0.039 and the character task r(10) = 0.651, p = 0.041 in the first session. We will discuss theories of hippocampal development and early word learning.

Authors

• Elliott Gray Johnson, University of California, Davis
  Presenting Author

• Ethan Fox, University of California, Davis
  Non-Presenting Author

• Lindsey Mooney, University of California, Davis
  Non-Presenting Author

• Simona Ghetti, University of California, Davis
  Non-Presenting Author