Guardian 'Brain Waves' project

Ulrich Pomper, Makoto Yoneya, Maria Chait

As part of the Guardian’s ‘Brain waves’ project we measured brain responses from singers Laura Mvula and Billy Bragg while they were listening to a self-selected, particularly emotionally meaningful song.

In a nutshell: We focused on the effects of music familiarity on the brain’s response. Our measurements demonstrate that both Laura and Billy's brains reacted very differently to the familiar and unfamiliar music. Brain responses and pupil dialation data suggest that Laura and Billy's brains distinguished the familiar from the unfamiliar snippets within 1/10 of a second. The unfamiliar snippets evoked more activity suggesting they required more 'brain energy' to process. These are observational data (no statistics have been performed) and must be verified using a larger group of participants.

To investigate Laura and Billy’s brain response to their chosen song we used two methods: EEG and Eye-tracking (see side-bar).

Billy and Laura being set up for the EEG experiments.

Because of the particular strengths and weaknesses of EEG and eyetracking, we decided to focus on the effects of music familiarity on the brain’s response. Will Laura and Billy’s brains react differently to familiar and unfamiliar music? How quickly will the differentiation happen?

For both Laura and Billy, we first we selected a second song which was unknown to them, but which was very similar in style, tempo, and instrumentation to the chosen song. We then cut both songs into very short snippets of less than a second (750 milliseconds) and then played back those snippets in random order. We were interested to see how fast their brain would differentiate between their favourite song which they know very well, and an unknown song which still sounded very similar. This may be akin to zapping through radio or TV programmes, immediately deciding on whether you want to stay and listen or move on.

Our findings:

The figures and results presented below are based on relatively little data, as well as representing single subjects rather that group averages over many subjects. While they are interesting and well in line with many previous studies, it is important to stress that the analyses are exploratory in nature. We are excited to investigate this systematically on a larger group of participants and will continue updating this page with data as it becomes available.

Brain activity measured from Laura and Billy showed the characterisitc pattern we normally observe in response to short sound snippets. However, the brain responses to the unfamiliar song appear larger (in amplitude) than those to the familiar song. This can mean that more nerve cells were active (in a more coherent and organized manner) when processing the unknown song. What was striking is how fast the brain responses differed between the two songs. The first peak showing a difference appears already a tenth of a second (100 milliseconds) after the snipped started, which is remarkable.

EEG results [1]:

Blue traces pertain to the familiar song; Yellow traces show responses to the unfamiliar song. The x axis in each plot represents time (0=the time at which the sound was played). The EEG data for both Laura and Billy reveal stronger responses to the unfamiliar song. Put differently, their brains recognized and differentiated the snippets from the familiar and unfamiliar song within a fraction of a second.

Likewise we see a much stronger change in the pupil size following the unfamiliar compared to the familiar song. This stronger response can be interpreted similarly to the EEG effects above: Surprise, novelty, and a violation of expectations. It might be that processing the new sounds in the unfamiliar song is more difficult and uses up resources in the brain, leading to a stronger pupil response. The familiar song on the other hand is very easy and straightforward to process, because it has been heard hundreds of times before.

Eye Tracking results:

Blue traces pertain to the familiar song; Yellow traces show responses to the unfamiliar song. The x axis in each plot represents time (0=the time at which the sound was played). The pupil dilation data reveal substantially larger responses to the unfamiliar song. These results mirror the pattern seen for EEG, above.

We also examined a second aspect of the EEG data: ongoing rhythmic activity. These ‘brain rhythms’ can be indicative of the general state of the brain. For example they differ strongly between being awake and fresh, being very tired, being asleep, or performing meditation. We see that a characteristic brain rhythm, the ,alpha rhythm’ at about 10 cycles per second, differs strongly between listening to the familiar compared to the unfamiliar song, for both Laura and Billy. If a part of the brain exhibits a strong alpha rhythm, this is generally interpreted as a sign that this part of the brain is disengaged and less active, compared to when it shows less alpha rhythm. The fact that we see less alpha for the unfamiliar song again suggests that both Laura’s and Billy’s brains are actively attending to the unknown sounds and trying to make sense of them.

Another potential interpretation is that listening to their preferred and highly familiar songs has a relaxing and partly disengaging effect on their brains. This fits with Billy’s feelings toward ‘Willin’, but potentially also with the fact that Laura’s song is a lullaby which she associates with a calm and relaxed state.

As a control we presented the stimuli used in Billy's experiment to 3 naive participants (who were unfamiliar with both songs). Their brain responses did not differ between the two songs suggesting that the effects observed above are indeed linked to familiarity and not due to other acoustic differences between the signals.

EEG results [2]:

Shown are the time-frequency representation of EEG activity between 1 and 20 Hz. Time is on the x axis (0=onset of sound snippet). The Y axis shows frequency content (‘brain wave’). Of particular importance is activity between 8-14 Hz, which is termed the ‘alpha band’ (marked with a rectangle in the figure). Strong alpha-band activity indicates a state of reduced awareness and attention. Conversely, reduced alpha-band activity indicates increased attention or more effortful processing. While the control subjects don’t show any difference between the two songs, reduced alpha-band activity is clearly visible in Laura and Billy’s response to the unfamiliar song. This indicate they are processing the unfamiliar song more attentively and/ or effortfully. Again, of particular interest is the fact that this difference emerges very quickly – within just a tenth of a second.

* Due to video time constraints the Billy video was focused on the fMRI results from our colleagues at Birkbeck.

EEG (electroencephalography) captures electric activity in the brain via electrodes placed on the scalp. This electric activity is temporally very precise and provides important information about when different computations are happening in the brain. For example, we can investigate the earliest time at which our brain can distinguish familiar or preferred sounds from unfamiliar or less preferred sounds. Due to distortions arising from the conductive properties of the skin, bone, and brain tissue, the EEG technique has relatively poor spatial resolution – it is not so great at telling where exactly in the brain the activation is coming from. For the same reason, EEG is better at picking up activity from the outer parts of the brain, such as the cortex. The cortex is heavily involved in processes such as perception from all sensory modalities, planning and initiating movements or abstract reasoning and planning ahead. These processes are well suited for an investigation with EEG. Other complex processes in the brain, such as emotions, motivation, or certain aspects of memory lie in deep brain structures and are more easily observed using techniques such as fMRI (functional magnetic resonance imaging).

Eye-tracking involves observing the various movements of the eyes using a very fast and accurate camera. Here, we are particularly interested in changes of the pupil size. A diverse array of processes can lead to increases of pupil size, such as a high attentive state, positive emotions, or performing difficult mental tasks.