Neural response to infant-directed speech: gamma band oscillatory activity in 4-month-old infants

Dublin Core


Neural response to infant-directed speech: gamma band oscillatory activity in 4-month-old infants


Marina Ciampolini




Infant-directed speech is an ostensive signal preferred by infants over adult-directed speech. We studied infants’ neural response to auditory stimuli by measuring gamma band oscillatory activity over the frontal area of the brain in response to ostensive infant-directed speech and non-ostensive adult-directed speech. Two groups of 4-month-old infants were presented with the same auditory stimuli, but the two groups differed in terms of visual stimuli (inverted vs. upright faces), being our study a part of a broader research project. We investigated only the auditory portion of the trial. We found that, in the inverted face group, the activation to the ostensive infant-directed speech was significantly enhanced while, in the upright group, this outcome was not found. These findings support the use of gamma band oscillations in assessing the basis of social communication and establish infants’ early specialization in understanding communicative signals directed to them.


Infant-directed speech; neural response; EEG; gamma oscillation


Experimental Design
We used data that had already been collected for a broader study, designed for the observation of the influence of auditory stimuli on face processing. In the main experiment, a total of 36 four-months-old infants was divided into two groups that differed on the visual stimulus presented at the end of each trial. Immediately after an auditory stimulus in IDS or ADS, the first group was presented with inverted faces, while the second group was exposed to upright faces. Participants in both groups were thus exposed to the same auditory stimuli, just before being presented to the visual stimuli, that were different depending on the group. In this research we focused only on the auditory portion of the trial, where participants were exposed to IDS or ADS (Fig. 1).

Figure 1. Representation of the complete trial presented to infants. In every trial, 18 infants were presented with upright faces, while the other 18 were presented with inverted faces. However, each infant was exposed to auditory stimuli in ADS or IDS, regardless of the visual stimulus. The green rectangle shows the portion of the trial analysed in this dissertation.
Infants were recruited from the Lancaster Babylab database. All were free of any known neurological, ocular or auditory abnormality and met the screening criteria of normal birth, born full term (gestational age >37 weeks), in the normal weight range (>2500g) and with an Apgar score of at least 8 at five minutes after the birth.
In our study we focused on infants’ neural response to the auditory stimuli (Fig. 1). However, the distinction between the two groups was preserved in order to observe possible differences between them. The group exposed to inverted faces included 18 infants (5 females, age range 117 to 161 days, M= 135.61 days). Thirteen additional infants were excluded owing to an insufficient number of artifact free segments (n=10), sleep (n=1), and technical issues during the experiment (n=2). The group presented with upright faces included 18 infants (5 females, age range 115 to 171 days, M= 145.22 days). 17 additional infants were excluded because of an insufficient number of artifact free segments (n=14) and technical issues during the experiment (n=3). In the final datasets (N=36) were included infants that provided artifact-free EEG recording in at least 10 trials within each experimental condition.
The auditory stimulus was the word “Hello” pronounced by a female voice using two different intonations: either IDS or ADS. The two words were recorded and edited with Audacity (v. 1.2.5) and Praat (v. 5.1) by setting a digitalization at 32-bit resolution and a sampling rate at 48 kHz. Both words were 850 ms long. The IDS stimulus had an average volume intensity of 61.86 dB, while the ADS stimulus had an average volume intensity of 61.50 dB.
Infants’ behaviour was video recorded for the entire duration of the test by a remote-control video camera placed behind the monitor. A pair of computer speakers situated behind the monitor were used for the presentation of the auditory stimuli. The infants’ EEG was recorded at a sampling rate of 500 Hz using a 124-channel Hydrocel Geodesic Sensor Nets (Electrical Geodesic Inc., Eugene, OR, USA).
Infants sat on their parent’s lap at a distance of 70 cm from a computer monitor. Each trial started with a dynamic fixation grabber at the centre of the monitor, for the duration of 2150 ms. Then the attention grabber stopped moving and the auditory stimulus (in IDS or ADS) was released by loudspeakers positioned behind the monitor and lasting 850 ms. The attention grabber remained still for an interval randomly varying between 200 and 400 ms. Then the grabber disappeared and the visual stimulus was presented for 1000 ms. A blank screen as an inter-trial interval between 1000 and 1200 ms was inserted between successive trials. Auditory stimulus in IDS or ADS were presented in a random order with the following constraint: no more than three successive trials of the same kind in a row. The trials were presented as long as the infants were willing to look at them. When they became fussy, the experimenters played a dynamic spiral together with an attractive sound. The session ended when the infant could no longer be attracted to the screen.
EEG measurement and data analysis
The electrical potential was band-pass filtered between 0.3-100 Hz. The filtered EEG was then segmented into epochs including 600 ms before stimulus onset and 1400 ms following the stimulus onset for each trial. EEG epochs containing artifacts caused by body and eye movement were automatically eliminated, whenever the average amplitude of a 80 ms gliding window exceeded 55 µV at horizontal Electrooculogram (EOG) channels or 150 µV at any other channel. In addition to automatic rejection, each individual epoch was visually inspected for further epoch selection. When <10% of the channels contained artifacts, the contaminated channels were replaced by mean of spline interpolation, while segments in which >10% of the channels included artifacts were rejected. Infants exposed to upright faces contributed on average 17.5 artifact free trials to the IDS condition (range: 8 to 36) and 18.34 trials to the ADS condition (range: 9 to 39). Infants exposed to inverted faces contributed on average 20.89 artifact free trials to the IDS condition (range: 8 to 38) and 20.78 trials to the ADS condition (range: 10 to 39).
In the artifact free segments induced gamma-band oscillations were uncovered through time-frequency analysis. These segments were imported into Matlab® and re-referenced to average reference through the free toolbox EEGLAB (v. The custom-made scripts collection WTools (available at request) was used to compute complex Morlet wavelets for the frequencies 10-90 Hz with 1 Hz resolution. A continuous wavelets transformation of single trials of EEG in each channel was performed, on 2000 ms long segments (600 ms pre-stimulus onset and 1400 ms after stimulus onset). The transformed segments were averaged for each condition separately. To remove the distortion in the time-frequency decomposition caused by convolution with the wavelets, 400 ms at each edge of the epochs were chopped, leaving a segment from -200 to 1000 ms around the auditory event. The average amplitude of the 200 ms pre-stimulus window was used as the baseline and was subtracted from the whole segment at each frequency.
Based on prior findings (Parise & Csibra, 2013), we selected the scalp location over the forehead (the average of channels 3, 9, 10, 15, 16, 18, 22, 23, corresponding to Fp2, Fpz, Fp1, respectively, Figure 2), a time window from 200 ms to 600 ms, and 25 to 45 Hz frequency window.
In order to verify that there were no significant differences between the accepted segment for each participant, a t-test between the average of accepted segments for each condition (speech) was performed and the same procedure was repeated for each group (face orientation). The mean amplitude was assessed by a repeated measure ANOVA with Speech (IDS x ADS) as a within subject factor, and Group as a between subject factor (upright x inverted). Paired-Sample t-tests were used for post hoc comparisons between the induced gamma-band oscillatory activity in response to IDS and to ADS. One-sample t-tests against 0 were used to assess whether the analysed gamma-band oscillatory activity differed significantly from the baseline.

Figure 2. Sensor layout for the Electrical Geodesics Inc. (EGI) 124-channel hydrocel sensor net, showing the locations of the electrodes under study (circled in green), averaged for measurement of the oscillatory activation.


Lancaster University


Excel files; Matlab files; SPSS files.




John Towse









Eugenio Parise

Project Level



Cognitive; developmental

Sample Size

36 infants

Statistical Analysis Type

Anova; t-tests



Marina Ciampolini, “Neural response to infant-directed speech: gamma band oscillatory activity in 4-month-old infants ,” LUSTRE, accessed January 25, 2021,