We constantly change our behavior between things like driving a car to chopping an onion. How does the brain switch quickly between different sensory-motor pathways to accomplish this?
One of the key problems in generating meaningful behavior is being flexible in routing activation from sensory inputs (such as the eyes) to the motor outputs (such as the muscles of your hand). For example, consider the situation of driving a car. You are driving on a highway, your hands are on a steering wheel and your brain sets up a effective sensory motor loop between what comes in through your eyes e.g., the distance from the edge of the lane and your hand movements; as the car slowly drifts toward one of the edges your hands make corrections and return the car back in the middle of the lane, where it should be.
Interestingly, you can do that quite effortlessly; you don’t need to think about what you are doing. You don’t need to pay a lot of attention, just a little bit. You don’t even need to make conscious decisions like “Now I will turn slightly right”, “Now I will turn slightly left”, “Oh, that was a bit too much left. Next time I’ll do a bit less of a correction.” Nothing of this sort is consciously happening in your mind.
Still, the driving works well. The car stays safely in the lane and you can even relax, sing along the music on the radio, have a conversation with a passenger next to you or just dwell on your thoughts.
But when we look what is happening behind this apparently easy job of keeping the car in a lane, it is quite miraculous. You were able to “re-program” your sensory-motor loops to do that job. Just by deciding “I will now drive the car and keep it in lane”, you effectively rearranged the flow of information from your retina, through the various parts of your sensory and motor cortices, ending in the muscles of your arms.
This is not a trivial achievement. The reason is that there are countless different ways to create sensory-motor associations. The particular task of setting the sensory-to-motor pathway needed to keep a car in lane is unique and fits exactly the needs of this situation. When you do a different thing—when not driving a car—you need different sensory motor associations. Even if you move from driving a car to driving a bicycle, a somewhat different sensory motor loop is needed. To steer a bicycle, different muscle movements are involved as compared to steering a car: the axis of rotation of the car steering wheel is different from the steering bar of a bicycle; the positions of the body are different as well.
But, while driving a bicycle is only slightly different from driving a car, there are other radically different activities. What about chopping onions, for example? Some people can chop vegetables rapidly by guiding the knife quickly up and down and in the same time drifting slowly sideways. When doing so, one has to maintain constant visual inputs in order to track where the knife is in relation to where the vegetable is and, most importantly, where the fingers are. Much like driving a car, onion chopping requires its own specific sensory-motor loop.
Switching between sensory-motor loops
A puzzling thing is that these two loops are considerably different; they differently map inputs to outputs. The two loops route sensory information through different pathways to achieve different types of control of muscle movements. A testament to their difference is also the fact that learning a skill of driving a car does not help with chopping onions and vice versa. Also, simultaneously chopping onion while driving a car would be quite hard, to put it mildly.
And yet, we are easily able to switch between these two sensory-motor loops. We just need to decide: Now I am keeping a car in a lane—and one sensory-motor loop is set in motion; Now I am chopping onions—and another sensory-motor loop is set in motion.
So, a fundamental question for brain science is: What does the brain do when it switches from one sensory-motor loop to the other? See an illustration of such switching problem in Figure 1.
Figure 1. An illustration of a dynamical routing problem. Sometimes it is necessary to route information from sensor on the left (e.g., a part of the sensory filed) into effector on the left side (e.g., an extensor of the left hand). Other times it is necessary to do just the opposite: route the information from left to right, as illustrated above. How can the brain flexibly achieve such dynamical changes in information routing?
The connectionist explanation
The traditional explanation tells us that the higher brain areas exert top-down modulation on the lower brain areas. The higher brain areas first make the decision on what to do and then inhibit and excite cells in lower brain areas as necessary to guide them to do accomplish the task.
Figure 2. The traditional, connectionist, approach presumes that dynamical routing is achieved such that higher brain areas modulate information flow of lower brain areas by “top down” inputs. In this example, inhibitory inputs are being sent to certain pathways (red) and not to others (grey). This way, information is made flow in a way that is illustrated in Figure 1.
For example, human frontal cortex may make a decision to “Go visit grandmother by car” and then, when the time comes to drive the car, the frontal cortex reorganizes the information flow from visual to motor areas of the brain, as illustrated in Figure 2.
Except that this does not agree well with empirical evidence. Brain structures can change behavioral modes without any impact of higher brain areas. People who have damage of high brain areas still switch their behavior. For example, unilateral spatial neglect is a common syndrome in which right hemisphere injuries to ventral fronto-parietal cortex lead to failure to pay attention to certain parts of the visual field. If asked to copy a drawing, such patients would only redraw the right half of the drawing, not even noticing that the left half is missing. On the face of it, such results seem to support the connectionist theory which assumes that the direction of attention is controlled by fronto-parietal cortex in a way illustrated in Figure 2
Foils of the connectionist framework
However, here is the foil. The connectionist framework predicts that if the input from right brain side higher areas is missing, there is no possibility to direct attention to the left. However, still with these injuries, the mentioned patients are able to fully direct attention to the left. It is only that it is very hard for them to switch from right to the left, but if they are directed to the left side by some salient stimulus (this could be a flashing light on the left, or simply a verbal instruction of another person “Did you forget something on the left?”), then they are just fine with perceiving everything from the left side. It is not that the higher brain area fully controls the direction of attention. It is more likely that the higher brain areas help push the tipping point of attentional switch (or fails to do so)—but the full tumble into a new focus of attention occurs independent of those areas (see e.g., Corbetta & Shulman 2011).
Also, other species that have nervous systems much less advanced than the human brain make the switches in behavior just fine; they alter behavior between feeding, mating, foraging, and even fighting ‘wars’ (as insects often do), all without a super-smart human-like frontal cortex. In fact, in majority of cases the switches in behavior occur without any cortex at all—just a few ganglia do the job. If that is not enough, consider the fact that the spinal cord itself can switch between different models of behavior e.g., from walking to running (Grillner & Rossignol 1978, Pearson 1993). See here a video of a decerebrated cat (a cat with its cortex removed) changing gait as the speed of treadmill increases.
see related post The Crisis of Computational Neuroscience
Fast adaptation as an explanation for sensory-motor switching
In my previous post I introduced the idea of fast adaptations as an interesting alternative to connectionism. Can this provide an alternative explanation that is compatible with the evidence that switching between different forms of behavior and hence, between application of different sensory-motor loops, occurs to a high degree locally—within the very circuits that execute the loop? The inputs from other “higher” brain areas may only be assisting the switch. They may be providing a suggestion of what kind of switching is needed. But the actual act of the switch itself, the temporary ‘re-wiring’, may largely be happening locally.
If neurons themselves ‘know’ when they should be responsive to inputs and when not and rapidly adapt, a neuron could individually reduce its responsiveness to (some of its) synaptic inputs. In this case there is no “central executive” that decides what should be done. Rather, the local circuits self-organize, forgoing the need to invoke higher brain areas.
Figure 3. Dynamical routing achieved by local adaptations, as would follow from the theory of hierarchical adaptations (aka practopoiesis). Red: the neuron has adapted to those inputs and is unresponsive. Gray: the neuron passes those inputs. Local adaptations produce the same routing effect as do the inhibitory inputs from the higher brain areas in Figure 2.
If some neurons adapt more and others less (or some parts of neuron adapt more and others less as in Figure 2) in response to incoming information, this will necessarily have an effect on the circuit as a whole such that the circuit will exhibit a specific type of behavior. Depending on the adaptation pattern, the routing of information throughout the network may go one way or the other. Moreover, a novel pattern of fast adaptations may “re-program” a network into doing something new, something it has not done before.
Adaptations, by their very nature, can also be described as attempting to achieve a certain goal. They have a function to optimize. Fast adaptations may thereby be the key mechanism that grants the brain its ability to flexibly respond to the demands posed by the environment on rapid time scales. Using fast adaptations, the brain could quickly move from state to state, from behavior to behavior, from sitting to standing, from grooming to mating, from chopping onion to driving a car. Stay tuned for future posts on possible mechanisms.
This is a part on the blog series on mind and brain problems by Danko Nikolić. To see the entire series, click here.
Pearson, K.G. (1993) Common principles of motor control in vertebrates and invertebrates. Ann. Rev. Neurosci. 16:265–297.
Grillner, S., Rossignol, S. (1978). On the initiation of the swing phase of locomotion in chronic spinal cats. Brain Res. 146:269-77
Corbetta, M., & Shulman, G. L. (2011). Spatial neglect and attention networks. Annual review of neuroscience, 34, 569-599.