Friday, January 23, 2009

Consciousness (4): Levelling with Mr B

Mr B has hypothesized that the brain is necessary and sufficient for conscious experience in humans (and probably other animals). The brain, unfortunately for Mr B, is an incredibly complex object. It consists of multiple interdependent processes that operate across different spatial and temporal scales.

Spatial scales in the brain
The left-hand side of the figure below lists neuronal processes that operate at different spatial scales (spatial scale increases as you go down the figure). Associated with each level of organization are experimental techniques typically used to access the phenomena at that level. A few of these techniques are listed to the right of each level.


Measurements at a lower-level are often used to help illuminate what is going on in higher levels. For instance, there are many studies that correlate single cell responses with behavior. A detailed anatomical characterization of individual cells slowly builds up a picture of the distribution and abundance of neuronal cell types in the entire brain.

While the techniques used at higher levels typically don't reveal the details of the lower-level processes (e.g., fMRI does not tell you what is happening in an individual neuron), the data from higher levels do provide useful clues about the functional roles of the lower level phenomena. For instance, behavioral studies can suggest how individual motor neurons help to govern behavior. This is important because, as we discussed before, part of Mr B's job as a biologist is to discover the function of the mechanisms he is studying. Just as you won't understand the biological role of sperm by focusing narrowly on how a sperm locomotes, it is not possible to understand the function of an individual neuron without studying its role in the neural network in which it is embedded, and ultimately the role of this network in behavior.

Temporal scales
As spatial scales increase, the relevant temporal scales also tend to increase. This is because the higher-level processes emerge from the interaction of many events at lower levels. For instance, a single action potential lasts about a millisecond, so network dynamics take place on longer time scales (network dynamics require presynaptic action potential propagation, neurotransmitter release, and postsynaptic responses often in a large number of neurons).

Note this positive correlation between spatial scales and temporal scales is not a hard and fast rule. There are lower-level molecular processes that can take longer than minutes to unfold, for instance.

At what level is consciousness?
At what level(s) of organization should Mr B begin his investigation of consciousness? This is not something that can be decisively answered a priori. He needs to dive in and do some experiments to discover the spatiotemporal organization of processes in the conscious brain. We will visit many of the techniques and levels of organization in the above chart as we follow Mr B.

However, Mr B knows enough neuroscience to form tentative hypotheses about the levels of organization required for consciousness. For instance, it is quite unlikely that a single neuron is sufficient for consciousness. This would be a brittle way to build an important process into the brain. This means that we should be looking at the level of the neural network or higher for the neural signatures of consciousness.

On the other hand, we know that the entire brain is not necessary for consciousness. People lose bits of their brain all the time (e.g., car accidents and strokes) without losing consciousness. Further, it is clear that behavior isn't constitutive of conscious experience. We can be paralyzed by curare but still conscious, and when we dream our motor system is effectively shut down but we still have experiences.

Hence, somewhere between small neural networks and areas/nuclei in the levels chart are the most likely candidates to find processes essential to consciousness. This suggests the time-scales of the processes should be relatively long (i.e., longer than the millisecond scale at any rate).

Minimal basis for consciousness
Mr B can use the above discussion to put a finer point on his working hypothesis from the previous post (namely, the brain is necessary and sufficient for consciousness in humans). Because the entire brain isn't necessary and sufficient, there must be some subset of the brain that is. It could be a certain set of cortical areas and subcortical nuclei. It could be the entire cortex, or perhaps just the brain stem. We just don't know right now. But what Mr B is after is this minimal subset of neuronal processes that is necessary and sufficient for consciousness.

Behavior and consciousness
Perhaps paradoxically, while Mr B doesn't think behavior is constitutive of consciousness, the most direct experimental indicators of consciousness he has are behavioral. That is, the best way to find out if someone is aware of something is to ask them if they see (or feel, or hear) it. There is presently no foolproof neuronal measure of conscious experience. While Mr B does think that such a measure should exist, it is something we will have to discover by doing the science.

The good news and the bad news for Mr B
That consciousness seems to be a higher-level phenomenon makes it both easier and harder to study. Mostly harder. The good news is that most of the techniques targeted at lower levels can be used to study consciousness, such as single cell recordings. And of course, those techniques developed to record larger-scale activity, such as fMRI, may also be revealing. This is useful because such techniques are noninvasive and can readily be applied to humans.

That consciousness is a relatively high-level phenomenon has an obvious down side: it is likely incredibly complicated. We don't even understand how the motor cortex controls limb movement, a relatively simple phenomenon. It will take a herculean effort over many decades to develop the empirical infrastructure required to establish a consensus about how the brain is conscious.

In the next post, I'll switch voices and then we'll start in on ambiguous stimuli.