The Limits of Imagination

Illustration depicting thought.

If you can dream it you can do it! Well, according to a recent study, maybe not so much, at least when your dream is to rehabilitate an arm weakened by stroke.

I have written several times on this blog about how imagery and visualization can be used to build coordination and physical skills. For example, it has been shown that imagining playing piano can make you play better, imagining strength training can make you stronger, and just watching sports can activate mirror neurons responsible for performing the movements you see. The basic idea is that thinking about a physical skill will activate almost the exact same neural pathways as actually performing it, so that you can better at something purely by visualization. What an optimistic "feel good" idea!

So I was interested to see a recent study supporting the pessimistic "feel bad" idea that stroke patients receive no apparent benefit from engaging in imagined movement of their weakened arm. I think this may shed some light on an important limitation in the usefulness of imagined movement.

The Study

The researchers recruited 121 patients who had, within the last six months, suffered a stroke causing significant weakness in one arm. The patients were divided into three groups.

The first group began a four-week program of imagined exercise. The exercises were all designed to activate the parts of the brain damaged by the stroke and responsible for arm movement. For example, patients imagined opening and closing the hand to activate their motor neurons. They watched videos of healthy hands to activate their mirror neurons. They also partook in some illusions - placing their weak hand under a video of a healthy moving hand. In total, they did nine hours of supervised imagery exercise and four hours of similar work on their own.

The second group didn’t do any motor imagery but instead spent their time visualizing flowers and other static scenes. (This is pretty much how I spend my time anyway.) The third group did standard stroke rehab. All three groups did standard physical therapy. After four weeks all groups showed similar improvements in the weakened limb. In other words, the imagery group had no advantage. I don’t have access to the full paper so I don’t know the interpretation provided by the authors. But here’s mine.

Starting from scratch versus brushing up

In some ways I am not that surprised. This is because there is a big difference between building a new skill from scratch and improving existing skills. My guess is that imagery is excellent for the latter task but probably close to useless for the former. In other words, imagery only works to optimize or modify existing skills, it can’t build entirely new ones. Because stroke rehabilitation likely falls into the category of building movement skills from scratch, the imagery didn’t help. Here’s a more detailed explanation of what I mean.

Imagine the process of learning a new language versus brushing up on a language that you haven't spoken for many years. In the case of brushing up, the neuronal connections representing the skill may be in storage, or rusty, or weak, but they are in there somewhere and can be brought back to full capability in a far shorter time than it took to build them in the first place. If you have a stroke that impairs movement of a certain area, regaining that movement is not like brushing up on an old skill, it's basically relearning the skill from scratch. The neuronal connections that supported the movement are not just in storage or a little disorganized, they are destroyed. Therefore, you need to find a completely different way to move - just modifying the old ones won’t help. I think this is a key point for understanding why imagery didn’t seem to help the stroke patients.

Reality versus virtual reality

Although imagining a movement and performing a movement are similar neurologically, they are quite different in one important respect - the movement offers sensory feedback from the real world. The feedback tells the brain whether the movement in question achieved its intended result. This is how we learn.

If you imagine movement, the only feedback you get is imagined feedback. And that might be of some value, provided you have enough experience with the movements in question to be able to accurately imagine their real world results. For example, I have enough experience with swinging a squash racquet to imagine the effects of all sorts of minor adjustments in my grip, or stance or swing path. So I can play around with my swing a little in my head and actually feel myself playing the game. And I would guess this makes me a slightly better player.

And even though I have never played handball, it has enough similarity to squash that I can imagine how to play, and I would guess I could get better at handball by just imagining playing it. But I would not get any better at doing back flips just by thinking about them, because I don’t have enough experience with this type of movement to have any idea at all what it would feel like. What would happen if I tucked my hips a little earlier, or threw my head back before my arms? I have no idea. I do not have enough real world data to run this virtual reality experiment in my head. And therefore I can gain no benefit from it. I think that is why the stroke patients could not benefit from just imagining movement of their hands. Their prior experiences are now irrelevant, as they depended on neurons that have been destroyed. Moving the hand is now a whole new ballgame.

Looking in the mirror

There is an interesting study by Daniel Glaser showing that your level of experience with a certain activity affects how your mirror neurons respond to watching the activity. Glaser asked some capoeira and ballet dancers to watch other ballet dancers and capoiera dancers do their thing. He found that the dancers had substantial activity in the part of the brain that controlled dancing when watching the form of dance they performed. In other words, when ballet dancers watched other ballet dancers, their mirror neurons lit up – when they watched the capoiera dancers … not so much.  The opposite was true for the capoeira dancers.

I would guess that the stroke patients didn't get much mirror neuron activity from watching other peoples hands or imagining their own because they had no preexisting maps to light up. Those maps had just been destroyed. They had to build new ones, not just light up the old ones.


So my guess is this - to build a road map for movement from scratch, you can’t just imagine the lines, you need to actually go into the world, drive on some roads, and take some notes. Once you have some virtual maps in your head, you can improve your navigation purely through imagination. But putting the maps down in the first place requires getting on the road.