Music machine

‘Shared music': Here’s a thought experiment that might be of interest to music therapists. The intention is to facilitate thought and discussion about what ‘shared music’ really is. It’s something I’ve often pondered, both from a music therapy perspective and as a jazz player. I also suspect that ‘shared music’ is a concept that is often used by music therapists as if everyone knows what it means. I’m not sure that I do. When we play music with other people, we often have the experience of feeling connected, as though we are inhabiting an alternative reality together. But could this be illusory? Music is after all a subjective experience. ‘Shared music’ sounds suspiciously like an objective concept. Look at it this way…

Imagine there was a machine that could improvise music. It could not only improvise, but respond sensitively and interestingly to other music. Now imagine that you were in a room playing music with a partition between you and the machine. Nobody told you about the machine, you simply heard the music it produced. Would you be able to tell that it wasn’t a real person making the music with you?

"This is moot point as such a machine is impossible anyway." Perhaps, but let’s pretend it is possible. The point of the exercise is to think about what it is about musical interaction that is human and which parts are ‘merely musical sound’. This is an important question for music therapists. Also there are lots of musical machines around now (auto-tune, samplers, loop pedals), even cheap and widely available ones, that would have been hard to imagine some years ago.

"Even if such a machine could be invented, it would be so complex that it would be easier for a real person to do the job anyway." This is an interesting one. In fact this is an issue that has been around for a while with regard to recorded music. The fact is that a lot of records which people enjoy are, to some degree, created by machines. The musical content may have been invented by a human, but the performance is electronic. Is this music necessarily ‘worse’ than music played on ‘real instruments’? The issue of improvisation and responsiveness complicates the issue, as a machine that can respond musically has to be a lot cleverer than one that simply follows a set of instructions, but it’s not impossible to imagine. However it is a valid objection in the ‘real world’. Studios still employ live musicians today, despite all of the advanced technology available, because there are some things that humans can still do better than machines. But this is a thought experiment and doesn’t have to take place in the real world.

"A machine, no matter how clever, could never achieve the subtlety of a live musician." This is perhaps getting near the crux of the matter. Maybe there are aspects of performance that are so subtle that we are not even fully conscious of hearing them, but which are nevertheless crucial to the ‘humanity’ of the music. Perhaps the tiniest of imperfections, perhaps subtle events that could never be anticipated are what differentiates human from machine. Perhaps there are musical things that only a real live person could ever produce which a machine, because of its essential ‘machine-ness’, could not. But then if this is the case, what are they? Because this is another important question for music therapists. It may be that the real question is ‘What frame of mind do we have to be in so that these essential human characteristics will happen as a matter of course?’ This is probably the approach that most music therapists, and indeed improvising musicians of any kind, take. Perhaps there might be a danger of being machine-like if we don’t get into this frame of mind. It’s also possible that a really great musician will always musically out-empathise a machine, just as a great chess grand master might beat the most powerful computer (though the machines have triumphed in that particular battle). However this might allow for a grey area where a lesser, even very good musician, or a great musician on an off-day, might not make the grade. Might they then be hard to tell from the machine? Or is it actually something essential that any musician, even an average one, possesses that differentiates them? Still we’re allowing for an imperfect machine with this reply. For the sake of argument, let’s pretend the machine is as responsive and inventive as the greatest musician who ever lived.

"There is something about the presence of a real person in the room, even if unseen, that is crucial to the musical interaction, some sense of another presence that, although not overtly part of the music, is vital to the experience of shared improvisation." This may well be true. Instinctively, I like this response. It allows for the possibility of this wonder-machine, but side-steps the issue. Basically, this is saying that there may be something almost mystical about the presence of another person in the room. Is this something that can be heard in the music? If not, how can you really tell whether the music is human or machine? Perhaps you can ‘feel’ the person’s ‘aura’. This answer is both irrefutable and also impossible to prove (unless we really had the machine to do the experiment).

"If you could invent such a machine, you couldn’t really tell the difference, but this is missing the point as shared music making involves other modes of simultaneous communication in order to be a fulfilling experience. The partition is the problem, as to be able to feel the musical connection you also need other ingredients such as eye contact, awareness of facial expressions and body movements, even extraneous noises like key rattles or intakes of breath." So perhaps musical connection cannot really be separated, particularly in a therapeutic context, from the whole experience of being in a room with someone else doing stuff. Maybe such a machine would be musically indistinguishable from a real person, but so what? It’s not just the music we’re interested in here, it’s the ‘whole experience’, which could mean many different things. Music itself is perhaps only part of the experience of making music with a person.

"When you improvise music with another person, you can just feel it. It’s a shared experience and that’s something that can’t be explained." Maybe this is the most honest answer. Unfortunately it doesn’t identify the essential musical ingredient(s) that separate this shared experience from the machine experience.

"Well maybe they’ll have been conned someone into believing they’ve been playing with a real person. They’ll be disappointed when the truth is revealed." So could there be something retrospective about shared musical experience? Could the reaction of the other musician, however small (jazz musicians getting together to play will often just nod after a tune, but that can be enough to say ‘I heard you’), be vital to the ‘shared-ness’. Anyone who’s ever done a background piano gig will testify to how dispiriting it can be to play to zero audience reaction. Even a single punter applauding out of pity can lift the spirits immeasurably, take it from me! So again we come back to the non-musical ingredients which can hold up a mirror to the shared music.

"If such a machine existed, it would be as good as a real musician. Let me know when you’ve invented one." Is this machine like ‘zombies’ in consciousness theory (hypothetical beings who behave exactly as we do, but have no self awareness)? You can imagine it, but even if it did exist, what’s the difference? The experience of the person playing would be the same with the machine as it would be with another person, but so what? The machine is emulating human characteristics perfectly, so the question we should really be asking is ‘what are those human musical characteristics?’ The question about ‘shared musical experience’ remains. Does it really exist, or can we only look at two separate experiences? Does ‘shared musical experience’ seem to imply an imaginary third party who is witnessing the sharing?

Afterthought: In my actual practice as a music therapist I believe I do experience connection to most clients through the music, but there remains something mysterious about this, and there are some who cause me to doubt the connection. We never know for sure what this ‘shared musical experience’ is like for the client. We work with at best partial knowledge, our own instincts and suppositions. Some clients tell us about the experience, but this is never really enough, and of course some can’t. Nothing is certain.

Benign Noodling

Knowing when not to play is an important preoccupation for me as a music therapist, and as an improvising musician. The client in a music therapy session needs to feel that there is space for them to play. It’s a fine balance, though, because sometimes they need some active music from me to get things moving. There can be times when I’ll think to myself – ‘Nothing’s happening. Let’s see what happens if I do this...’ and often the results are surprising. In one session a child who was playing tentatively, with no clear pulse, began to play the drums with a strong groove once I introduced a regular pattern on the piano. Was I pushing them away from their own music into something of mine, imposing my agenda? I don’t think so. It felt more like I was opening a door, and they were choosing to walk through it. The music felt genuinely enlivened and positive and this was reflected in the child’s comments afterwards.


I’m in the habit of trying to find psychodynamic formulations for things that happen in music therapy. I know there’s a long running argument that says you don’t need to do this, that music can be therapeutic on its own terms, but I think it helps me to avoid taking certain things for granted. For example, if I make some music with a client in a session and it feels good, I’ll try to think about why this is. Is it for the right reasons? Because it might be that the music feeling good just shows that I’ve successfully imposed my own music, and the client is being compliant, identifying with my projection. It’s worth at least considering this possibility. Perhaps psychodynamic thinking is partly about not entirely trusting your instincts, but taking the time to question your unconscious motives. It’s difficult though, because this could equally stymie a healthily developing musical relationship. I might over-think. And part of the trick with improvisation is to learn to trust yourself, to feel rather than think your way through the music. Hopefully a lot of the thinking has been done before, during the long process of developing competency on an instrument and finding a voice of your own. So when thinking about music therapy, there’s an interplay between thought and feeling. The process for me goes something like: have a feeling about the music, think about what it means, test the thoughts by experiencing how they feel, get back with the flow. This process can take place within very short spaces of time, seconds even, or can be part of the longer process of therapy, including reflection in supervision.


Sometimes a client will spend some time just fiddling around on an instrument, with no particular feeling of direction or development. Electric guitars seem to stimulate this for some reason. There’s a fascination with the sound that you can get with very little effort. Twang the strings with the distortion turned up and it sounds pretty cool. There can be quite long periods of time during which the pure enjoyment of this cause-and-effect sends the client into a kind of trance, staring into space as they pluck the strings and feel the vibration coming from the amplifier. A ‘psychic retreat’ perhaps? Using the music to create distance between the client and the world? Perhaps this is the ‘doubling of the self’ described by Mahns in Analytical Music Therapy, the rock guitar equivalent of whistling in the dark. Whatever, I feel like I don’t want to interrupt. The client disappears into the experience of the moment and it’s time for me to disappear too. Or maybe I’ll do some aimless noodling of my own at the piano or the drum kit, but nothing that gets in the way too much. Music is supposed to mean something, traditionally, but sometimes a bit of meaningless meandering has an important function. For a little while, nothing matters. The music therapy space becomes an empty space. These are either moments in which nothing is being achieved, or possibly they are the most important moments in therapy, but I’m not sure which. Nothing is being achieved – the double meaning is striking. If therapy is a space to take a break from the expectations of the outside world and just be, at least some of the time, then nothing might be quite an achievement.