The brain as a network – a new hypothesis?
August 29, 2010 3 Comments
The BBC News Sci/tech section recently ran a few stories about neuroscience and technology that I considered to be a bit oversold. There was one article in particular I wanted to write about: Brain works more like internet than ‘top down’ company. The article refers to a recent study in PNAS that used injections of dye to trace connections between brain areas. The article claimed that it demonstrated that the brain is an interconnected network like the internet, counter to the apparently prevailing view of the brain as a hierarchy. The technique could also apparently lead to a map of the brain. Now, I was initially going to take the BBC to task for what seemed to be another oversold and rather general piece of research. However, after having a read through the paper, it seems that the authors may, to some extent, have oversold the novelty of the implications of the work, and the groundbreaking-ness of their technique. The line of reporting by the BBC only compounded the problem.
Before I start, I’d like to say bravo to the BBC for actually getting on board with linking to the actual research paper behind a story, rather than just linking to the journal front page and trying to claim that was doing us some sort of favour.
The study demonstrates a new anatomical tracing technique by investigating the connectivity of the dorsomedial nucleus accumbens, a region in the brain that is thought to be involved in appetite, stress and clinical depression. The area was chosen because the authors’ believed that its wide range of apparent functions suggested that it had a more complex pattern of connections with other structures in the brain than had been previously thought. The technique – double coinjection (COIN) network tracing strategy – involves making two injections of mixtures of tracers. The mixtures are made up of an anterograde tracer, which travels forward to areas of the brain that the injected area makes connections to, and a retrograde tracer, which travels backward to areas of the brain that make connections with the injected area. The two injections of tracer mixtures then allow a hypothetical circuit of four different structures to be investigated.
Example labelling, and diagrammatic illustration of the double coinjection technique.
The authors report that the tip of the dorsomedial nucleus accumbens is part of a complex looped circuit, including regions of the brain involved in stress, depression and the control of feeding and metabolism. As the study involves several injections of anterograde and retrograde tracers, the results are quite extensive. Briefly, the tip of the dorsomedial nucleus accumbens (ACBdmt) is one node in a four node closed loop, consisting of the anterior region of the lateral hypothalamic area (LHAr) and a region of the stantia innominata which receives connections from ACBdmt (SIdmtr); anterior regions of the paratenial and paraventricular nuclei (PTa/PVTa); and the infralimbic cortical area (ILA). The closed loop features feed-back connections and connections out to other brain regions. This loop is embedded in a series of other loops with their own feed-back and feed-forward connections.
Connectivity of the tip of the dorsomedial nucleus accumbens
The body of this study is an investigation of a previously unanswered neuroanatomical question to demonstrate a new technique. The major finding is which regions of the brain are connected to which, and how. However, it seems to have been pitched in both the original paper and in the BBC article as a new development in neuroanatomy that opens up possibilities that were not previously available, and that the features of the circuitry support a ‘proposed alternative’ conception of brain connectivity. I don’t believe is it quite such a ground breaking new tool, nor has it told us new information about how the brain in general is connected.
Neuroanatomists have long used anatomical tracers to investigate connectivity of the brain. The use of multiple injections to label several elements of a circuit simultaneously is also something that anatomy studies have done for years. The use of mixtures of tracers also doesn’t seem to be the technique introduced in this study. Rather, it seems to be an improvement on current methods by combining of several existing tools. The double COIN technique compresses what would be several separate studies of several connections into one. I’m not saying this isn’t a good thing; on the contrary, this technique will reduce the number of animals needed for a study, as well as saving researchers’ time and money. I just don’t know if it warrants a fanfare for the reasons the authors and the BBC seem to have given. It doesn’t seem to me to herald the dawn of a new ability to describe the connectivity of the brain that we haven’t had before. It’s also not going to be easy to make a “full map of the nervous system”, regardless of the technique used.
The other major point the article was sold on is that it demonstrates a looped, networked style of connection in the brain, with feedback and cross communication between regions. The study authors contrast this view with the hierarchical view of brain connectivity, where input from independent regions funnels up to a central processing unit, then filters back down again. Now I think it’s fair to describe this as a 19th century view of brain connectivity, but I don’t think that it is fair of the authors to say that this is the way that brain connectivity is “usually described”, or that neuroscience is dominated by a top-down/bottom-up linear flow view of the brain. The ideas of feedback and feed-forward mechanisms within a pathway, and interaction between disparate regions of the brain are not new, and this paper is neither the first to propose nor demonstrate a networked view of the brain.
Thompson, R., & Swanson, L. (2010). Hypothesis-driven structural connectivity analysis supports network over hierarchical model of brain architecture Proceedings of the National Academy of Sciences DOI: 10.1073/pnas.1009112107
About Neuromancy
Maybe I haven’t been in the field long enough, but I haven’t heard any neuroanatomist talk about neural networks as if they believed in a strictly (or even predominantly) hierarchical model of their function. It does seem a bit like the authors of this paper have missed the boat, in that sense. It is a neat set of techniques, but I also don’t see this work as more than a small step-wise improvement on/development of pre-existing tracing techniques.
Also, neat blog. It’s good to see other new-ish bloggers – it makes me feel less out-classed!
Yeah, I didn’t really get their point about networks. There are probably a few subsystems, which, if you look at them in isolation, could be considered hierarchical, but I’d bet that pretty much every system in the brain is considered to have some feedforward/back mechanisms and crosstalk at most levels, even if there are levels which you could consider to be ‘low level’ and others to be ‘high level’. When I first thought about writing the post I was worried it was the BBC’s misunderstanding, but they do seem to be pitching the paper as evidence in support of a new alternative way of thinking about the brain.
Also, I’m very envious that you’re studying cephalapods and neuroscience, I sometimes wish I’d come to neuroscience from pure biology. It’s nice to have other bloggers commenting here too!
Cephalopod neuroscience is actually a tangential interest of mine. The research I’m involved in (and hope to continue doing) is in animals models of cocaine addiction, stress, and reward – much more behavioral neuroscience-typical. I came to cephalopods through neuroscience, which I came to through psychology, initially.