But since I was one of the prime offenders with the explanatory flowchart, and I seem to be using them more and more, I wanted to offer a stalwart defense of the use of flowcharts and diagramming in neuroanthropology, especially as both contribute to the practice of partial explanation. So, to pick up a theme from a number of my posts, ‘yes-you’re-right-but-I-still-disagree,’ here’s why I find flowcharts particularly useful and think anthropologists should be doing a lot more diagramming to highlight complex patterns of causation, situating more broadly the parts of complex systems that they are exploring.
But before I go any further, I need to direct all our readers to the recent announcement of the first Neuroanthropology conference which Daniel posted. Although I want to post, I feel like I also want to keep drawing attention to this announcement. But on with it…
As with all of her comments, I felt that Prof. Quinn cut to the quick, highlighting an issue in a cautionary fashion rather than rejecting specific arguments our panelists were making (at least I don’t think she was just calling me out…). In the case of flowcharts, Prof. Quinn suggested that diagramming relationships was a preliminary step, not a final goal – at least that’s one of the ways that I took her comments – and I agree.
I haven’t done anything at all to confirm whether or not there’s a trend, but I do know that, certainly in cultural anthropology, we don’t often make diagrams or draw pictures any more, in spite of our interest in Great Diagrams in Anthropology. It had certainly been years since I attempted to sketch anything before some of the conference workshops asked us to try.
Quinn’s comment is ironic as one of the discussion points in a conference I just attended in Scotland at the University of Aberdeen was that drawing as a practice, actually sitting down and making a picture with pencils or charcoal, has been disappearing from our field. Fewer and fewer anthropologists, some participants asserted, use sketching or other visualizations as a technique within anthropological research or theory-making.
My own growing use of diagrams, grounded in a dynamic systems approach to neural variation, as I look back over my posts, is definitely trending steadily upward (see, for example, talent diagrams and balance diagrams). Although I take Prof. Quinn’s cautionary note to heart, something like a flowchart, I believe, can help us to see what portions of emergent systems we are examining in any single research project or discussion.
The diagramming process can be a way of acknowledging and preserving the centrality of causal processes that are not themes in our work. A good diagram can exceed our narrative, showing the threads in a dynamic system that we are not tracing at this moment, in a sense acknowledging visually the limits of our own theoretical narratives.
The use of such unresolved diagrams can help us to avoid the tendency to see different emphases in explanation as inherently antagonistic, heading off unnecessary and pointless debates (shades of, ‘It’s the genes!’ ‘No, it’s the environment!’ ‘Nature!’ ‘Nurture!’).
In other words, causal diagramming can aid tremendously in what we could call partial explanation: proposing inherently and self-consciously incomplete causal narratives in complicated, emergent systems. I believe that, in any complex system, the researcher will be aware of contributing variables that cannot be fully encompassed in the current project or even by their academic discipline. With important contributing factors beyond the scope of the project, a partial explanation is both the only possible outcome, as well as a potentially helpful contribution to the overall, communal effort to explore a complex system, especially if it highlights a neglected set of dynamic relations. (Don’t worry, this is vague, but I’ll get really concrete in a minute).
In contemporary cultural and cognitive anthropology, many theorists are either uncomfortable with any sort of causal explanation or working with surprisingly simple causal assumptions, even if they may not be explicit about it, or even aware of it. For example, certain post-structuralist positions assume that, in terms of social efficacy, ‘power causes everything’ – although you’d never say it that clearly. When pushed to talk about causes for things like psychological variation in individuals, I know I used to get pretty evasive, or offer a whole list of likely contributing factors, without really attempting to clarify how those factors might be related, weighted, interacting, checking or negating each other, or otherwise concretely linked to the ‘outcome.’ Similarly, talk about ‘mutual causation’ or cycles of reiterative causation still assume relationships even though the discussion can often stop there, without specifying in any greater detail.
In fact, trying to describe causal relationships in polypotent (or multi-causal) systems can lead to two opposed unproductive tendencies: simplistic reduction to a single scale or reduced set of causes in what is really an emergent and multicausal system; or a non-systematic piling up of causal relations until outcomes look either impossible to predict or inevitable and over-determined (but usually only in retrospect). In other words, sorting out and even visually representing causal processes may actually be more important when we are not dealing with a simple chain of one-way interactions, although it’s hardly easy.
Nested relations and neuroconstructivism
One excellent example of the use of flowcharts to sort out mutual causation and interrelations between processes on different scales appears in Westermann et al. (2007), an article on neuroconstructivism that I discussed in an earlier post. While traveling through London a couple of weeks ago, I managed to get ahold of the two volume collection on the same subject; I may eventually write more about the two books soon, although I found that the first chapters of the first volume were pitched at a frustratingly general or introductory level (possibly because I’m not a member of the intended audience).The flowcharts in Westermann et al. (2007), however, highlight extremely well the partiality of an explanation, demonstrating both that the immediate process under discussion is part of a more complex network and that processes at one scale or level of resolution are embedded within larger-scale developments as well as contingent upon smaller-scale events unfolding.
In a recent presentation to the Macquarie University Centre for Cognitive Studies, I offered several modified versions of Westermann and colleagues’ diagram in order to discuss where I thought ‘cultural’ and ‘social’ factors entered into the dynamic relations that they describe and how these sorts of considerations might be operationalized as research-ready factors (instead of left vaguely as just ‘culture’ or ‘society’). Of course, I was talking about sports, and I didn’t mean to be exhaustive, but I thought that the diagram worked out nicely.
I added in the red arrows on the original Westermann and colleagues diagram to suggest feedback effects of physical training in sports, focusing on some of the possible causal dynamics that were not highlighted in the original diagram (and article). The blue-ish box tries to add in a few of the elements on the cultural and social level that will also feed into the individual-level developmental dynamics, for example, how what I’ve called ‘health ideologies’ create exercise patterns (everyday behaviour) that are socially modeled and taught. These influence an individual’s body behaviours, but they don’t determine them; we all know lots of examples of people going to exercise classes or sports training sessions and not doing the techniques the same way. In this case, the impact on the nervous system is going to be different than with another variant of the ‘same’ activity.
One thing the modified diagram highlights is that ‘culture’ and ‘society’ themselves are aggregate dynamic systems that need to be resolved into constituent processes and factors. As I was doing this diagram, I was reminded of the sorts of systems thinking of someone like Talcott Parsons, which is hardly surprising as there is a lot of shared ground between contemporary dynamic systems theory and the sorts of cybernetic systems theory that probably influenced Parsons’ own modeling of social systems. (No, I haven’t yet pulled out Parsons again to think through him again — hey, I’ve got conferences to help organize, a farm to look after, and just spent three weeks on the road in Europe. I’ll get to it…)
But another thing that the Westermann and colleagues diagram suggests is the difficulty involved in skipping intervening levels of analysis. For example, in some discussions of the neurological roots of religious thinking, or in evolutionary contributions to behaviour, there may be a tendency to gloss over intervening scales of interaction between the level of the brain region or even the neuron and the level of institutionalized religious tradition or the evolutionary time-scale phylogeny. The point is not that there’s no relation between these levels of phenomena, but rather that there are emergent processes at each level between them, which intervene and contribute to the final outcome.
After all, with processes as complex as human cognitive capacity or behaviour patterns or social psychological development, it’s very clear to all but the most militant reductionists that any comprehensive theory is going to need to take in factors beyond the expertise of any one researcher, even any one intellectual discipline. This doesn’t mean that specialized knowledge is any less valuable, only that we need to have a clear sense of how partial explanations fit together and a way of acknowledging other causal relations without becoming bogged down in the complexity.
Dynamic systems modeling & partial explanation
Often in sciences, physics is held up as the ideal for explanatory elegance, rigorous description of interactions that can be rendered as mathematical relations. Of course, biological, psychological, and social phenomena — let alone ecological and meteorological systems — are notoriously resistant to this sort of rigorous reduction. But it’s also important to find manageable ways to grasp complex systems, so we don’t just give up. I was inspired to use graphic modeling as a thought tool for staking out systems more comprehensively than I could investigate or explain by dynamic systems theorists, especially Peter Taylor.
Peter now teaches in the Critical and Creative Thinking Program at the Graduate College of Education of the University of Massachusetts Boston. Originally from Australia, Peter trained in ecology and science and technology studies before moving into promoting reflective practice and training teachers. He has published widely on complexity, including his book, Unruly Complexity: Ecology, Interpretation, Engagement (Chicago, 2005; the table of contents and prologue are available here, a summary here).
I met Peter when I spent a year at Brown University on a post-doc, participating in a seminar on embodiment in which Peter was an active player (thanks to Anne Fausto-Sterling for the opportunity). One of the things that Peter was so good at was using visualization strategies to combine what initially appeared to be opposing perspectives on core issues. I found this synthetic approach more rigorous than many of the synthetic theoretical work I had encountered before meeting him, particularly amenable to dealing with complicated human-environment relations.
Peter has explored environmental phenomena that he characterizes as having ‘unruly complexity or “intersecting processes” that cut across scales, involve heterogeneous components, and develop over time.’ In his writing, he has used a range of diagramming techniques. For example, he offers the following diagram in a discussion of soil degradation in Mexico, suggesting the deep historical depth and complex relationships that caused the environmental degradation (see Taylor 2001).
The goal of this sort of modeling is not to reduce artificially or ignorantly the complexity of a phenomenon until our explanation is empirically inadequate but logically elegant, nor is it to reproduce the territory with the map, offering empirically rigorous but utterly incoherent explanations. As Peter Taylor writes of his complex diagrams and explanatory links (and see the original for some great case studies, including a really interesting discussion of women with depression):
The strands, however, are cross-linked; they are not torn apart. In this sense, the account has an intermediate complexity — neither highly reduced, nor overwhelmingly detailed. (Taylor 2001: 318)
One of the salutary effects of these sorts of diagrams of ‘intermediate complexity’ is that they tend to highlight the incompleteness of our own explanations, going beyond what any one scholar or scientist could be expected to grasp. Self-conscious partial explanation produces space for collaboration and engagement rather than falsely pitting researchers investigating different strands of a phenomenon against each other. Being overly ambitious with a partial explanation, assuming that one has The Theory instead of a theory among many, can put researchers in conflict who should be cooperating.
To go back to the red arrows and additional boxes I added to the Westermann and colleagues diagram, the diagram makes obvious that the causal dynamics I am focused upon are nested within a tangle of additional relations. A person would have to be particularly obtuse to disregard the partiality of any subset of the relations included in the diagram.
But the practice of diagraming more comprehensively and expansively than we might thoroughly discuss in our writing and explanation also wards off some of the least generous criticisms of our work, for example, the assertion that a person who focuses on social or cultural influences in psychological development necessarily must be ignoring genetic factors.
I agree with Naomi Quinn, that a flowchart is not a theory. In fact, a flowchart might even be adequate to link up a number of theories, illustrating their complementarity and intersection by acknowledging each theory’s partiality.
Taylor, Peter. 2001. Distributed Agency within Intersecting Ecological, Social, and Scientific Processes. In Cycles of Contingency: Developmental Systems and Evolution. Susan Oyama, Paul E. Griffiths, and Russell D. Gray, eds. Pp. 315-332. Cambridge, Mass: MIT Press.
Westermann, Gert, Denis Mareschal, Mark H. Johnson, Sylvain Sirois, Michael W. Spratling and Michael S.C. Thomas. 2007. Neuroconstructivism. Developmental Science 10(1): 75–83. doi: 10.1111/j.1467-7687.2007.00567.x (pdf available here)