Reading-induced shifts of cortical speech representations in dyslexic and typically reading children

Full text

Reading-induced shifts of cortical speech representations
in dyslexic and typically reading children
Linda Romanovska, Roef Janssen, Milene Bonte
Dept. Cognitive Neuroscience, Maastricht University, Maastricht, The Netherlands
Background and Introduction
One of the proposed mechanisms underlying reading problems in dyslexia is impaired letter-speech sound integration [1]. We pursue the hypothesis that
this impairment stems from reduced auditory cortical plasticity and test the hypothesis in 8-10 year old dyslexic and typically reading children using
(f)MRI and text-based recalibration, a short-term audio-visual learning paradigm [2,3].
In this paradigm, an ambiguous speech sound between /aba/ and /ada/ is combined with disambiguating text to bias the perception of the speech sound in
subsequent auditory-only post-test trials. Previous research in adults reveals that typical readers show a shift in the perceptual boundary of the ambiguous
sound towards the text thus showing a recalibration effect whereas this shift is not observed in dyslexic readers [4].
Discussion
The behavioural results show that text-based recalibration shifts speech perception in both dyslexic as well as typically reading children. Contrasting the
cortical activation maps of the two groups reveals more activation within brain areas associated with more effortful reading [5] in dyslexic children and
more activation within the auditory cortex and ventral higher-order visual areas in typically reading children. These results not only suggest
different mechanisms of reading-related audio-visual learning in developmental dyslexia but also highlight the importance of considering the dynamic nature of
reading development when investigating group differences between typical and dyslexic readers.
References
[1] Blomert (2011). NeuroImage, 57(3), 695-703. [2] Keetels et al. (2016). Atten Percept Psychophys, 78(3), 938-945. [3] Baart et al. (2012). Acta Psychol, 140(1), 91-95. [4] Keetels et al. (submitted). [5] Dehaene-Lambertz et
al. (2018). PLOS Biology, 16(3), e2004103.
Correspondence to:
Linda Romanovska
linda.romanovska@maastrichtuniversity.nl
https://mbic-languagelab.nl/
Department of Cognitive Neuroscience
Faculty of Psychology and Neursocience
T +3143 388 1478
Maastricht University
P.O. Box 616
6200 MD Maastricht, The Netherlands
(x4) Instructions AV AV A R A (x8)
aba +
ITI = 2s ISI = 8-12s 1-2s
Figure 1: Each recalibration run consisted of 6 mini-blocks comprised of an exposure block and post-test
trials. AV=audio-visual exposure trials; A=auditory-only post-test sound; R=response trial
Behavioural results
Intriguingly, both dyslexic as well as typically reading children showed a
behavioural recalibration effect (Figure 2 below).
(f)MRI results
The overall cortical activation during the exposure blocks overlaps with key areas previously associated with reading, however dyslexic readers show
less activation within the auditory cortex compared to typical readers (Figure 3). Contrasting the activation maps of the two groups reveals differences
in cortical activation patterns. Whereas the dyslexic readers exhibit more activation within parietal and frontal areas, typically reading children engage
auditory cortex and occipito-temporal/fusiform areas more (Figure 4).
0
0.25
0.5
0.75
1
A?A-1 A?A A?A+1
Test Sound
0
0.25
0.5
0.75
1
A?A-1 A?A A?A+1
Proportion /aba/ responses
Test Sound
ABA exposure
ADA exposure
Figure 4: Contrasting activation maps between dyslexic and typically reading
children.
Typically reading children
Dyslexic children
FDR<0.01
t(14709) 8.0 3.0 -3.0 -8.0
Typical readers Dyslexic readers
Methods
Participants: 15 dyslexic (9±0.73) and 38 typically reading (9±0.78) children were recruited from local schools and a specialised dyslexia institute. Here we include preliminary data of 12
children from each group.
Imaging: (f)MRI data were collected on a Siemens 3T Prisma MRI scanner. The recalibration task was acquired in four 5min. functional runs (2,5mm3, TR = 2000 ms, TA = 1100 ms, TE = 35.8
ms, 50 slices). Anatomical T1 images were collected using an MPRAGE sequence (1mm3, TR = 2300 ms, TE = 2.98 ms, 192 sagittal slices).
Data anlysis: The preprocessing of the anatomical and functional data was performed using Brainvoyager QX V2.8. Individual cortex surfaces (N=24) were reconstructed from grey-white matter
segmentations and anatomically aligned to a group-average target cortical representation based on curvature. Subsequent functional analyses were performed at the surface level.
Figure 3: Overall cortical activation during the audio-visual exposure
trials across groups.