The Sounds of Science

The use of headphones is just one of many aspects of cycling behaviour which generates lively debate; but debate which rarely braves the cold world of data and quantitative study, preferring to inhabit the opinion columns of newspapers and the soundbites of politicians.

Let’s try and move it on a little.

Signal-to-noise ratio

Much of the discussion of headphones is, sadly, noise. This reached its deafening peak in November 2013, in the wake of a spate of cycling deaths in London. Most notably Boris Johnson publicly decried the use of headphones, even though there was no evidence that they were even in use by any of the deceased, let alone that they had contributed to the collisions.

On the one hand this sparked keen defence of headphones by those who use them, while on the other, the tabloids wheeled out the “zombies” invective, which the Guardian questioned.

So, who’s talking sense?

Is riding with headphones safe?

Often, there’s a simple question we need to answer. But it’s extremely rare that the real world to which they pertain is anything like so simple, even when it might initially appear to be.

So, I think it’s reasonable to translate “is riding with headphones safe?” (and that should really be the subtly but importantly different “does riding with headphones introduce an unacceptable level of danger?”) into the following questions:

1. Is the rider’s ability to hear external sounds diminished by the presence of headphones and/or the addition of music, speech etc?

2. Is the resulting loss of ability to hear external sounds of a degree and nature which diminishes their ability to build a dynamic mental picture of the environment and traffic around them?

3. Is the rider’s concentration diminished by the additional sounds of music, speech etc?

4. Is the rider’s loss of concentration of a degree and nature which diminishes their ability to concentrate on the environment and traffic around them, and/or their ability to control their own cycle?

5. Are there any strategies the rider can readily adopt which can sufficiently mitigate any additional risk that may arise from above?

6. Does familiarity with the use of headphones reduce some or all negative effects that might be experienced at initial use?

Logical consistency

Before we start actually answering those questions, let’s take a quick look at them in terms of their logical implications.

Firstly, one has to consider deaf and partially deaf people. Logically, if one was to argue that the loss of ability to hear sounds was in itself sufficient reason to prohibit cycling with headphones, one would also have to argue that deaf and partially deaf people would have to be banned from cycling.

Secondly, one has to consider cars. Cars have stereos in, and therefore, logically, if one was to argue that loss of concentration through listening to music or speech was in itself sufficient reason to prohibit cycling with headphones, one would also have to argue that car stereos would have to be banned.

So, whilst the questions above are well worth asking, if the answers to them are to be used as arguments in favour of prohibiting the use of headphones there are some pretty hefty logical implications if that argument is to avoid hypocrisy.

But let’s take a pop at answering them anyway.

Experimental data

So, why does the discussion rarely include data and research? Well, it’s partly because there’s not much of it about. But there is some.

Testing ability to hear

Let’s kick off with a paper by de Waard et al, which studied cyclists’ response to an auditory “stop” signal whilst using headphones or mobile phones. It was a small study, of 25 participants, and it found that

Cycle speed was not affected by listening to music, but was reduced in the telephone conditions. In general the response to auditory signals worsened when participants listened to music, in particular when listening with in-earbuds loud auditory stop signals were missed in 68% of the cases. However, when listening with only one standard earbud performance was not affected.

This is hardly surprising: listening to music does reduce the rider’s ability to hear other sounds, and in-ear headphones markedly increase this effect.

Of course, this begs the question: When does one hear an auditory stop signal in real-world road use? I think the use of “a stop signal” carries unfortunate connotations of an on-carriageway directive, and therefore assumptions that there is an inability to respond to “a stop signal”, rather than “an auditory command”. Stop signals are not auditory. They are red lights, white lines, and so on. All visual.

Sadly, this is the most-cited piece of research around, and it’s very often misinterpreted in that exact way.

Comparisons with driving

Some Australians also did a pertinent experiment. It’s not a peer-reviewed paper in a scientific journal, but it is a fairly conscientious test that makes quantitative and qualitative measurements. It found that

A bike rider with ear-bud earphones playing music at a reasonable volume hears much more outside noise than a car driver, even when that driver has no music playing.

A bike rider with in-ear earphones playing music at a reasonable volume hears about the same outside noise as a car driver with no music playing, but more than a car driver playing music.

This reinforces the point made earlier, that if one is to argue for prohibiting cyclists from using headphones, one has to also argue for the prohibiting drivers from using stereos and even windows in order to be logically consistent.

Research overview

The Netherlands’ Institute for Road Safety Research (SWOV) mentioned the de Waard et al paper in a summary of research activities. In it (see p8) they make a notable inference from the experiment:

An experiment in which the participants had to cycle while they were listening to music and using their phones, and were told to stop when they were given an audio signal, indicated that their cycling became less safe than when they did not listen to music.

This is a little disturbing from a research institute, in assuming that some loss of hearing ability (which, crucially, is what de Waard et al tested) is inherently less safe. It’s one of those things that seems obvious, but may or may not be true.

And it’s also the exact type of unfounded inference that I believe de Waard’s “stop signal” test encourages.

On the same page, the authors note that

The use of these types of devices while cycling appears to increase the risk of being involved in a crash.

Where they use the phrase “these types of devices”, they are making a common mistake, which is to gather mobile phones and music players into the same category. They are markedly different: the former require interaction and are widely shown to introduce significant cognitive distraction.

The mistake can be highlighted by the data they cite:

15% of the Dutch cyclists said they listened to music during (almost) every trip.

More than 3% of the cyclists report making or receiving a phone call during (almost) every trip.

Approximately 9% [of] bicycle crashes with injury are preceded by the cyclist using devices.

These data are wholly consistent with the hypothesis that phones cause significant added risk and music players do not. Without being able to separate these two groups, however, it actually seems quite possible that music listeners may be under-represented in the crash figures: are they actually less likely to be involved in a collision?

Is there even a worthwhile hypothesis that music listeners are safer? Just as there is a hypothesis that non-helmeted cyclists may be safer because they are less inclined to risk compensation, is there also one that aurally-restricted cyclists are safer because they do not rely on sound signals (which are unreliable) and instead increase their reliance on visual checks (which are highly reliable)? This takes question 5 above and changes it: Do the strategies adopted by headphone users actually improve safety?

Some soundbites

When we break the problem down into testable questions, we find some interesting things, both in terms of the logic and the statistics. Two conclusions stand out:

• Given sound measurement data and the sameness of the auditory content, it seems impossible to argue a case for banning headphones without that argument being wholly inconsistent with the use of roads by deaf people and/or people in enclosed vehicles.
• The statistics relating to collisions are murky. There seems to be no compelling evidence that safety is compromised, despite there being clear evidence for loss of hearing. Moreover, the data may even show an improvement in safety and one might hypothesise that this is because loss of hearing increases reliance on the more reliable sense of sight.

However, whatever the data show, like any discussion about restricting cyclists’ behaviour on the basis of things that seem obvious to others, I’m sure this will rumble on.

And, like all those discussions, all this noise achieves is distraction from the real way to solve the problem.

Hit play, hit repeat.