Journal Article•DOI•

Dynamics of brain activation during learning of syllable-symbol paired associations

Q: What are the contributions mentioned in the paper "Dynamics of brain activation during learning of syllable-symbol paired associations" ?

In this paper, the authors examined the long-term effects of audio-visual learning using transcranial direct current stimulation.

Jarmo A. Hämäläinen¹, Tiina Parviainen¹, Yi Fang Hsu², Riitta Salmelin³•Institutions (3)

University of Jyväskylä¹, National Taiwan Normal University², Aalto University³

01 Jun 2019-Neuropsychologia (Pergamon)-Vol. 129, pp 93-103

TL;DR: The results show that the short-term learning effects emerge rapidly (manifesting in later stages of audio-visual integration processes) and that these effects are modulated by selective attention processes.

read less

Abstract: Initial stages of reading acquisition require the learning of letter and speech sound combinations. While the long-term effects of audio-visual learning are rather well studied, relatively little is known about the short-term learning effects at the brain level. Here we examined the cortical dynamics of short-term learning using magnetoencephalography (MEG) and electroencephalography (EEG) in two experiments that respectively addressed active and passive learning of the association between shown symbols and heard syllables. In experiment 1, learning was based on feedback provided after each trial. The learning of the audio-visual associations was contrasted with items for which the feedback was meaningless. In experiment 2, learning was based on statistical learning through passive exposure to audio-visual stimuli that were consistently presented with each other and contrasted with audio-visual stimuli that were randomly paired with each other. After 5–10 min of training and exposure, learning-related changes emerged in neural activation around 200 and 350 ms in the two experiments. The MEG results showed activity changes at 350 ms in caudal middle frontal cortex and posterior superior temporal sulcus, and at 500 ms in temporo-occipital cortex. Changes in brain activity coincided with a decrease in reaction times and an increase in accuracy scores. Changes in EEG activity were observed starting at the auditory P2 response followed by later changes after 300 ms. The results show that the short-term learning effects emerge rapidly (manifesting in later stages of audio-visual integration processes) and that these effects are modulated by selective attention processes.

...read moreread less

Summary (3 min read)

Jump to: [1. Introduction] – [2.1 Experiment 1] – [2.1.2 Stimuli and experimental design] – [2.1.3 Data recording and analysis] – [2.1.4 Statistical analysis] – [2.2.1 Participants] – [2.2.3 Data recording and analysis] – [2.2.4 Statistical analysis] – [3.1.1 Behavioral results] – [3.1.2 MEG results] – [3.2 Experiment 2: Passive learning] and [5. References]

1. Introduction

Relatively little is known about the immediate learning processes in the human brain that occur at the beginning stages of training of cross-modal associati ns.the authors.
While studies examining the long-term learning effects have been important in establishing the brain mechanisms involved in cross-modal processing, it is not known which of these brain mechanisms are used during the initial steps of the learning process, and if there are distinct stages of learning during which some of the mechanisms are more important than others.
Studies on long-term effects of audio-visual learning provide a starting point for expected short-term learning effects.
The superior temporal sulcus in the left hemisphere has been implicated particularly in processing of well-established letterspeech sound combinations, thus mostly reflecting lo -term audio-visual memory representations (Raij et al., 2000; van Atteveldt et al., 2004; Hashimoto & Sakai, 2004, M AN US CR IP T AC CE PT ED Blomert, 2011).
Both active and passive tasks were used to examine possible general neural mechanisms related to learning of audio-visual associations.

2.1 Experiment 1

Thirteen adult participants were included in the analyses (26.3 years on average, range 21-38 years; 7 female, 6 male; 12 right-handed, 1 ambidextrous based on self-report).
From the total of 15 participants, one participant was excluded due to magnetic artifact from a tooth brace and one due to excessive eye blinks during the visual stimulus presentation.
None of the participants had lived in Japan or studied Japanese (relevant for the choice of visual stimuli, see below).
The study was approved by the Ethics Committee of the Aalto University.

2.1.2 Stimuli and experimental design

Auditory stimuli were recorded by a female native Finnish speaker in a sound-attenuated booth.
The delayed audio presentation was introduced in order to allow a clean access to cortical processing of the visual symbol without contamination by auditory activation, motor response, or response error monitoring.
Accuracy and reaction time (with respect to question mark onset) were obtained for each trial.
For thispurpose two categories of trials were created, learnable and non-learnable .
For the other half of the symbols the participants received the word ‘incorrect’ as the feedback and thus their association to syllables could not be learned (non-learnable category).

2.1.3 Data recording and analysis

MEG data was collected using a 306-channel (102 magnetometers, 204 planar gradiometers) whole-head device (Elekta Oy, Finland) at the MEG Core of Aalto NeuroImaging, Aalto University, Finland.
The head position was monitored continuously using 5 small coils attached to the scalp (3 on the forehead and 2 behind the ears).
Noise covariance matrix was calculated from the baseline interval of the averagd responses.

2.1.4 Statistical analysis

Repeated measures ANOVAs (category [learnable, non-lear able] x quarter [1st, 2nd, 3rd, 4th] x hemisphere [left, right]) for each time window and region of interest were conducted.
Effects involving interaction between category and quarter were of interest.

2.2.1 Participants

Seventeen adult participants were included in the analyses (26.2 years on average, range 20-35 years; 14 female, 3 male; 16 right-handed, 1 left-handed based on self-report).
The study was approved by the Ethics Committee of the University of Jyväskylä, Finland.
Each experimental trial started with a fixation cross shown at the centre of the screen for 745 ms.
To examine the effect of association learning, two categories of trials were created, learnable and non-learnable.
Half of the visual stimuli were always presented with its corresponding auditory stimuli (earnable category) while the other half of the visual stimuli were randomly paired with three auditory stimuli (non-learnable category).

2.2.3 Data recording and analysis

EEG data was collected using a 128-channel NeurOne amplifier (Bittium Oy, Finland) with Ag-AgCl electrodes attached to the HydroCel elctrode net (Electrical Geodesics Inc., OR, USA) with Cz electrode as the reference.
Electrode impedance was checked at the beginning of the recording and aimed to be below 50 kOhms for all channels.
The data was analysed using BESA Research 6.1 (BESA GmbH, Grafelfing, Germany).
EEG was first examined for channels with poor data qu lity (mean: 4, range 0-10) that were rejected at this stage, and then segmented into trial-based time windows of -200 - 700 ms with respect to the visual symbol onset (200 ms pre-stimulus baseline).

2.2.4 Statistical analysis

EEG data was then examined using cluster-based permutation tests (Maris & Oostenveld, 2007) in BESA Statistics 2.0.
After initial t-test comparison between conditions of interest, the results were clustered based on time points and channels.
Significance values for the clusters were based on permuted condition labels.
Cluster alpha of 0.05 was used with 3.5 cm channel neighbor distance and 3000 permutations.
The learnable and non-learnable conditions were compared in each block.

3.1.1 Behavioral results

All participants were able to learn the correct audio-visual associations during the first half (1st and 2nd quarters) of the MEG recording with only a few errors made after that.
Accuracy was scored based on the response to the question “do the symbol and syllable form a pair” (for non-learnable items the correct answer was ‘no’).
The mean accuracy rate was 90 % and 93 % and mean reaction times were 436 ms and 513 ms for the learnable and non-learnable categories, respectively, across the whole training session.
There was a clear effect of training in the accuracy and reaction time measures with improving performance towards the end of the session as shown in Figure 5.

3.1.2 MEG results

The MEG data showed clear visual and auditory evoked fields .
The response was similar for the two categories during the first quarter of the session, started to differ between categories during the second quarter, and remained different btween categories until the end of the session.
The distributed source analysis paralleled the sensor level trends.
Activation loci were found in the left and right inferior temporo-occipital areas as well as left frontal areas and right central-parietal areas in the time window of the slowly growing difference between the categories .
This was due to a decrease of source strength from the first to the second quarter.

3.2 Experiment 2: Passive learning

Similarly to the active learning experiment, the EEG data for the passive learning was examined in four blocks of equal length (10 min).
There wno statistically significant condition differences (p = 0.274) in the ERPs measured during the first 5 minutes whereas the between-category differences during the second 5-minute sub-block were statistically significant (cluster 1, p < 0.045 at 165-276 ms, fronto-central distribution) .
The authors expected to see learning effects at the early sensory responses as well as in later time window linked to perceptual learning and audio-visual integration in brain areas that previous studies have linked to short-term cross-modal learning (e.g., Raij et al., 2000; Hashimoto & Sakai, 2004).
Frontal cortices also showed enhanced activity bilaterally after 10 minutes of training.
The time window after 300 ms matches well with the current active learning task and with earlier EEG studies examining audio- M AN US CR IP T AC CE PT ED visual learning using a congruency manipulation (Shams et al., 2005; Karapidis et al., 2017; 2018).

5. References

Audiovisual integration of letters in the human brain.
The grey box represents the approximate time window for the difference between the stimulus categories given by the cluster-based permutation satistics.

Did you find this useful? Give us your feedback