The article by Max Bertolero and Danielle S. Bassett tried to sell us on something called "network neuroscience," a term that hasn't been around for many years. The article is behind a paywall, but in other articles that you can freely read online, we can read about the claims of people who are adherents of this academic discipline. For example, in the article here ("Inside the Network Neuroscience Theory of Human Intelligence") and the article here ("Network Neuroscience Theory of Human Intelligence" by Aron K. Barbey) we can read theorizing similar to that in the Scientific American article.
At the top of the "Network Neuroscience Theory of Human Intelligence" article by Barbey, we have a claim that "general intelligence, g, emerges from individual differences in the network architecture of the human brain." This claim makes no sense. Let us consider a single human being in isolation, not considering any other humans. There has to be some reason why this particular person has intelligence, understanding and consciousness. It makes no sense to claim that his intelligence arises from differences between his brain and the brains of other people. If all of the people in the world were destroyed by some giant asteroid colliding with Earth, and a single astronaut was left in a space station, then there would be no brain differences between this person and other humans; for there would be only one human left. But the surviving human would still be intelligent. So what sense does it make to claim that intelligence arises from "individual differences" in brains?
This seems to be the general approach of adherents of "network neuroscience":
(1) They start out by claiming or insinuating that individual parts of the brain can be strongly associated with particular mental functions.
(2) They do various forms of "network analysis" using the mathematics of network analysis that has grown up over recent decades, mainly in reference to computer networks. In their analysis, regions of the brain are considered as nodes in a network.
(3) These thinkers claim that something in their network analysis provides insight as to how brains could think or understand things.
There are several serious problems with this approach. The first is that there is no good evidence that particular parts of the brain are causes of particular mental functions relating to thinking or understanding. In his paper Barbey states, "Converging evidence from resting-state fMRI and human lesion studies strongly implicates the frontoparietal network in cognitive control, demonstrating that this network accounts for individual differences in adaptive reasoning and problem-solving – as assessed by fMRI measures of global efficiency and structural measures of brain integrity." The term "frontoparietal network" basically refers to the front part of the brain. This kind of claim that thought comes from the front part of the brain is debunked in my lengthy post "The Dubious Dogma That Thought Comes from the Frontal Lobes or Prefrontal Cortex," which includes links to many neuroscience papers.
When making the claim I just quoted, Barbey gives references for several neuroscience papers. One of those papers is a 2010 paper by Barbey and two others, entitled "Dorsolateral prefrontal contributions to human intelligence." That paper found an average IQ of 91 for some 19 patients who had lesions in the dorsolateral prefrontal cortex. But the study here with a much larger sample tells us that 37 patients with damage to the dorsolateral prefrontal cortex had an average IQ of 97.4, only very slightly below average (Table 1 and Table 2). There is no truth to insinuations that we can tell from fMRI studies that some particular part of the brain is responsible for some intellectual function. Contrary to the misleading visuals so often given by neuroscientists, it is not at all true that particular parts of the brain are activated far more strongly than other regions when some particular intellectual task is done (outside of the occipital lobe used for vision, the differences tend to be only about a half of one percent, about what we would expect from random fluctuations).
In the article Barbey starts talking about small-world networks. The wikipedia.org article on small world networks describe them as a type of network in which "most nodes can be reached from every other node by a small number of hops or steps."
A small-world network
Barbey tries to persuade us that a brain is a small-world network. This analysis is incorrect. The only natural and straightforward way to analyze the brain from a network perspective is to consider individual neurons as the nodes of the network. Considered in that way, the brain is not at all a small-world network. As this paper says, "If considered at the cellular level, brain networks are also unlikely to form classical small-world networks." The actual number of hops needed to travel from one end of the brain to another is in the hundreds or thousands, and it is not at all a small number such as five or six.
Figure 12 in the paper here gives a graph comparing connection probability between two neurons and distance. The graph shows that at a distance of 500 microns, this connection probability falls to essentially zero. Now, a human brain is 15 centimeters across, which is 150,000 microns. Such a distance is equal to 300 lengths of 500 microns. So we can very roughly calculate that the number of hops to get from one side of the brain to another is 300 or more. Considering such facts, we cannot at all judge the brain to be a small-world network, which can be traversed by fewer than 10 hops (as in the visual above).
So how do people such as Barbey state claims that the brain is a small-world network? Rather than judging a neuron to be a node of a brain network, which is the natural way to consider things, they artificially and arbitrarily declare certain regions of the brain to be the nodes.
Barbey states this:
"Recent advances in network neuroscience further elucidate the functions afforded by a small-world architecture, motivating new insights about how brain network topology and dynamics account for individual differences in specific and broad facets of general intelligence, represented by the Network Neuroscience Theory."
Claiming that the brain is a small-world network is erroneous, unless you decide that the nodes of a brain should be some regions of the brain that you have arbitrarily selected, rather than a neuron, which is the real natural node of a brain. Even if the brain was a small-world network, or any other type of network, that would not clarify how a brain could generate a thought or a concept or a belief or some understanding of something.
In Barbey's paper there is a diagram with the strange title "Hierarchical structure of general intelligence." There is actually nothing hierarchical or structural about general intelligence. It is an unstructured thing without anything like the parent-child relations that are found in hierarchies.
After defining an ICN as an "intrinsic connectivity network," Barbey towards the end of his paper tries to put his theory in a nutshell. He states, "In summary, Network Neuroscience Theory proposes that general intelligence depends on the dynamic reorganization of ICNs – modifying their topology and community structure in the service of system-wide flexibility and adaptation." This has no weight as an explanation for how a brain could produce a mind, or how neurons could produce thought or understanding. I could give countless examples of cells below the neck that dynamically reorganize, and people don't believe that such cells cause thought. There is no reason why we should think that some reorganization effect in a brain can explain thinking or intelligence or understanding. Moreover, there isn't very much physical evidence for reorganization effects in a brain (although there are many cases of minds rebounding after severe brain damage, there is little physical visual evidence of brains restructuring to achieve such a rebound).
The Scientific American article "How Matter Becomes Mind"
by Bertolero and Bassett offered similar content that did nothing to explain how it is that a brain could produce a thought, an idea, or some understanding of something. The article had some dubious poorly substantiated claims that the brain consists of modules. It sounded somewhat like the discredited old theory known as phrenology, which is described in a wikipedia.org article as being "based on the concept that the brain is the organ of the mind, and that certain brain areas have localized, specific functions or modules."
The authors say, "It is the music your brain plays that makes you you." If that were true, then each day you would be a different you, because each day your brain "plays different music" by transmitting different electrical signals. But you stay you with remarkable constancy from year to year, despite such daily fluctuations in "brain music." And in many cases, a heart of someone will stop, and their brain will "stop playing music" as its electrical activity very quickly ceases. But in such cases a person will often continue to have vivid experiences (what are called near-death experiences). So you can't be some "music" your brain is playing.
After making the statement above, the authors went on and on with a musical analogy, continuing it for several paragraphs. Previously the authors had referred to "massive orchestras of neurons." Now they referred to "musical compositions" played by brain modules, they referred to "the symphony in our head," and they referred to "the brain's music." This is all worthless for explaining how matter could give rise to mind, or how a brain could produce an idea, or how neurons could produce a thought. Music is a sound, not a thought, an idea or understanding. If if were true that the brain is constantly "playing music," this would do nothing to explain how matter could give rise to mind, or how a brain could produce an idea, or how neurons could produce a thought. A modern computerized keyboard can be set on a continuous loop, to continuously play music; but the keyboard doesn't have mind or thoughts or understanding to the slightest degree.
The Scientific American article by Bertolero and Bassett refers us to some brain scanning data they used. They state, "To better understand what was happening, we used publicly available data from a landmark study known as MyConnectome, in which Stanford University professor Russell Poldrack personally underwent imaging and cognitive appraisals three times a week for more than a year." They seem to have drawn a conclusion (about re-routing of brain connections) based on some brain scans of a single individual, and that's kind of like drawing a conclusion about baseball hitters based on how often your spouse gets a base hit.
Connections and networks do nothing to explain cognition. Let us imagine a System X which is a dense network of 100 trillion nodes (each consisting of a chip or transistor). Imagine that in System X each node is connected to a billion other nodes. In this System X the number of nodes is greater than the number of cells in the brain; and also the number of connections per node is many thousands of times greater than in the brain. But despite all this network connectivity which is so much greater than in the brain, there would not be the slightest reason for thinking that this System X would be capable of having a thought or an idea or an actual understanding of something. See my post "Physical Connections Do Nothing to Explain Cognition" for more on the futility of trying to use connections or networks to explain the human mind.
In the articles I have mentioned, there is some reference to some studies that try to show that some small fraction of intelligence can be predicted by some analysis of brains or brain networks or brain connections. Such studies are typically dubious for a variety of reasons. The scientists involved in the studies will typically be free to play around with any of hundreds of different ways of analyzing brain scans and crunching the fMRI data, until they find one that may seem to predict a small fraction of intelligence. If one such way is found, it is not very impressive; for we might expect them to get a little success on some tiny fraction of the analytic permutations, purely by chance.
Consider two theories. The first (call it Theory A) is that your mind is produced by your neurons and the connections between them. The second (call it Theory B) is that your mind is not at all produced by your brain, and that your mind is an aspect of your immaterial soul. Under this Theory B it might be that the brain acts as a kind of valve to limit your mind. Without our brains we might have minds fit for godlike thoughts and brilliant cosmic contemplation, but our brains may limit our minds so that we feel comfortable doing mainly the tiny little chores of earthly living. Under such a case, it is quite possible that brain parameters might affect intelligence, for they might affect how much of a "valving effect" occurs to limit your intelligence. So if you were to show some limited relation exists between brain states (or brain wiring) and intelligence, you do nothing at all to show that the brain is the source of the human mind. Such a finding would be equally compatible with Theory A and Theory B. A study purporting to give such a finding would neither show how matter could produce a mind, nor would it actually support the claim that matter does produce a mind.
As if to try to compensate for the weakness of their explanations, the article made sure to use a gigantic font for the title "How Matter Becomes Mind." I've never seen a font so big in a magazine; the font was so big the title took up half of the page. I offer this advice to scientists: don't so often claim to understand things you do not, and if you make such claims, don't state them in a gigantic font.
No comments:
Post a Comment