Infant brains detect arithmetic errors

Abstract

A current debate is whether increased looking time in infancy is related to violation of expectations. In this study, 6- to 9-month-old infants' brain activity was analyzed during presentation of correct and incorrect solutions to simple arithmetic equations [(e.g., presentation of 1 + 1; one doll on a TV monitor, with another doll added from behind a screen, followed by a solution of 2 (correct) or 1 (incorrect)]. Infants looked longer at incorrect solutions than at correct ones. Event-related potentials, time-locked to the presentation of the solution, also differed between conditions, with greater negative activity for the incorrect solution condition. Spectral analysis showed a similar pattern to that of adults observing correct and incorrect arithmetical equations. These findings show (i) that the brain network involved in error detection can be identified in infancy and (ii) that this network can support an association between looking time and violation of expectations.

Fig. 1.

Stimuli were modeled after those used by Wynn (1). The puppet versions of the different mathematical equations (1 + 1 = 1, 1 + 1 = 2, 2 − 1 = 1, and 2 − 1 = 2) were videotaped and displayed on a 21-inch flat-screen TV. (Upper) Stimulus phases from the infants’ point of view. (Lower) Stimulus phases from the video-camera point of view, facing the infants’ faces for looking-time calculations, with the mirror behind the child indicating the stimuli being displayed.

Fig. 2.

Topographical maps showing the voltage distribution and the localization of the effect on the scalp in the different conditions and groups. (Upper) Infants. (Lower) Adults. The correct solution condition is shown on the left, and the incorrect solution condition is shown on the right. The channel group that was used for the analyses of both infants and adults is marked in white and localized over the mediofrontal area of the Geodesic 128 channels, localized approximately between and including electrodes Cz and Fz of the 10–20 system.

Fig. 3.

Grand-averaged ERP. Notice negative voltages for the incorrect condition compared with the correct one (between 330 and 530 ms).

Fig. 4.

Time frequency analysis results. (Top and Middle) Grand-averaged time–frequency plots from the Fcz electrode. Dark areas indicate low-power values, and light areas indicate high-power values. (Left) Data from infants. (Right) Data from adults. (Top) Relative powers of brain activity after the presentation of a correct solution. (Middle) Relative powers of brain activity after the presentation of an incorrect solution. Note the relative increases in the power of δ-, θ-, and lower α-frequency bands for the incorrect condition compared with the correct one. (Bottom) Corresponding ERP wave graphs. The time window is shown in gray.