Saturday, November 10, 2012

Can causality be opposed to explanation?

Greetings from Toronto, where I gave some talks and had a great time with the archaeologists at the University of Toronto.

Actually, I don't particularly like epistemology.

I just read American Anthropologist’s 2012 review article on archaeological publications in 2011 (Hauser 2012). I came away confused about author Mark Hauser’s epistemology. I can’t seem to translate his approach into my own understanding of social science concepts of explanation and causality. Like many archaeologists, Hauser seems reluctant to discuss these things clearly and explicitly, leaving readers to puzzle them out from fragmentary and cryptic phrases.

I was particularly confused by two statements in Hauser’s paper:

·       “addressing long-standing questions about agriculture with new data require a shift from causality to explanations of process in specified contexts.” (p. 185)

·       “One major shift was a general move from the search for causes to explanation of cultivation, domestication, and farming.” (p. 186).

I find this puzzling, because to me, explanation means finding the cause for something. So what does Hauser mean by a shift from causality to explanation? His phrasing suggests that causality is the old and bad way of looking at things, while explanation is a new and good approach. This statement might give a clue:

·       “interpretive narratives find their expressions in situated explanations—a grounding in the messy idiosyncrasies of evidence—context—culture—history that run counter to more ambitious inclinations to craft explanatory models of history at larger scales.” (p.184)

Leaving aside the confusing conjunction of four words with hyphens, this statement suggests that Hauser is advocating explanation at a small scale. But does he mean a smaller spatial scale (we should concentrate on explaining a particular event in a particular place, not a large-scale spatial process like an empire or world system)? Or does he mean a smaller analytical scale (we should explain this particular household, and not households in general)? In my view, we need to explain things at a variety of spatial and analytical scales, and it doesn’t make much sense to prefer one level over another. It depends on the research question one is asking.

To me, talk of explanations at different scales suggests the concepts of proximate and ultimate causality. These concepts were first articulated in biology by Ernst Mayr (Mayr 1961). Proximate causes concern immediate factors such as ontogeny, and ultimate causes produce evolutionary explanations. While new work in biology has complicated this dichotomy, the basic distinction remains important in that field (Laland et al. 2011).
The proximate-ultimate distinction in causality is also important in the social sciences:

·       “ultimate explanations are concerned with why a behavior exists, and proximate explanations are concerned with how it works.”(Scott-Phillips et al. 2011:38)

Some writers parallel the approach of Laland et al in biology by advocating a more complex concept of social causation, but retaining the basic insight of the proximate/ultimate distinction. John Gerring (2005, 2012), for example, talks about “causal distance”: how far is the cause from the event it generates? Lieberson and Lynn (2002) use the phrases “underlying conditions” and “precipitating conditions” for these concepts.

But I don’t think this is the kind of thing Hauser has in mind, since he seems hostile to (or at least dismissive of) the concept of causality.

OK, let’s step back and look at what kinds of explanatory models are available in the social sciences (and, by extension, in archaeology). I always return to Charles Tilly for this kind of thing. I’m not a very good abstract thinker, and Tilly clarifies many epistemological issues in terms I can understand. According to Tilly, five explanatory strategies are available in the social and historical sciences. The following is my paraphrasing of: Tilly (2001:365) and Tilly and Goodin (2006:12-13); see also Tilly (2008). The text following *** are my own comments.

1.     Skepticism. The world is too complex to explain. *** I think this would be the strong postmodern position.
2.     Law-seeking accounts. Social life is said to exhibit empirical regularities that at their highest level take the form of laws; explanation then consists of subsuming particular cases under broadly validated empirical generalizations or even universal laws. *** This  is “logical positivism,” a framework associated with Carl Hempel. Binford and the new archaeologists promoted this approach to explanation, EVEN THOUGH IT HAD ALREADY BEEN DISCREDITED FOR SOCIAL SCIENCE BY PHILOSOPHERS OF SCIENCE! This was one of the biggest mistakes of Binford and the new archaeology, and it set archaeological epistemology back for decades!
3.     Propensity accounts. Social units are seen as self-directing, whether driven by emotions, motives, interests, rational choices, genes, or something else. Explanation then consists of reconstructing the state of the social unit—for example, an individual’s beliefs at a given point in time and space—and plausibly relating its actions to that state.  *** Most examples of propensity accounts follow methodological individualism, or the standard model in economics that says social phenomenon can be explained by the goals, decisions, and behavior of individuals. But other individual-level approaches, such as phenomenology, probably fit here as well.
4.     Systemic explanations. Particular features of social life are explained by specifying their connections with putative larger entities: societies, cultures, mentalities, capitalist systems, and the like. Explanation then consists of locating elements within systems. Functional explanation is a subcategory of systemic explanations. *** This includes both cultural explanations and explanations invoking large structural features such as social classes or world systems.
5.     Mechanism-based accounts. This approach claims that explanation consists of identifying in particular social phenomena reliable causal mechanisms and processes of general scope. Causal mechanisms are events that alter relations among some set of elements. Processes are frequent (but not universal) combinations and sequences of causal mechanisms. *** This is the way I think about explanation and causality. To explain an event is to identify the mechanisms responsible for bringing about that event. There is a BIG literature outside of anthropology on causal mechanisms. Some good starting points are:  (Bunge 2004), (Hedström and Ylikoski 2010), (Sampson 2011), and various works by Tilly, e.g. (Tilly 2008). Or check out Daniel Little’s blog, “Understanding Society” and search it for mechanisms.

When I started thinking about Hauser’s puzzling statements on causality and explanation, I thought Tilly’s scheme of the five explanatory approaches would clarify the situation, but now I’m not so sure. I still can’t figure out what Hauser’s explanatory position is, or how he can oppose the terms causality and explanation. I have almost given up trying to understand archaeological writing in this genre, that might be called “postmodern-light.”  I guess I will just have to admit defeat here.

My broader point is that archaeologists need to discuss epistemology more frequently, more explicitly, and more in tune with the relevant social science literature. Our own field has a rather poor track record in this area, and cultural anthropology is not much better. If this stuff is new to you, check out Tilly or Bunge or some of the other sources below. When I find myself in a conceptual difficulty, I often ask myself, “what would Tilly say about this?” Maybe you should ask that question too.

Bunge, Mario
2004    How Does It Work?: The Search for Explanatory Mechanisms. Philosophy of the Social Sciences 34(2):182-210.

Gerring, John
2005    Causation: A Unified Framework for the Social Sciences. Journal of Theoretical Politics 17:163-198.

2012    Social Science Methodology: A Unified Framework. 2nd ed. Cambridge University Press, New York.

Hauser, Mark W.
2012    The Year in Review, Archaeololgy: Messy Data, Ordered Questions. American Anthroologist 114(2):184-195.

Hedström, Peter and Petri Ylikoski
2010    Causal Mechanisms in the Social Sciences. Annual Review of Sociology 36:49-67.

Laland, Kevin N., Kim Sterelny, John Odling-Smee, William Hoppitt, and Tobias Uller
2011    Cause and Effect in Biology Revisited: Is Mayr's Proximate-Ultimate Dichotomy Still Useful? Science 334:1512-1515.

Lieberson, Stanley and Freda B. Lynn
2002    Barking up the Wrong Branch: Scientific Alternatives to the Current Model of Sociological Science. Annual Review of Sociology 28:1-19.

Mayr, Ernst
1961    Cause and Effect in Biology. Science 134:1501-1506.

Sampson, Robert J.
2011    Neighborhood Effects, Causal Mechanisms and the Social Structure of the City. In Analytical Sociology and Social Mechanisms, edited by Pierre Demeulenaere, pp. 227-249. Cambridge Universitiy Press, New York.

Scott-Phillips, Thomas C., Thomas E. Dickins, and Stuart A. West
2011    Evolutionary Theory and the Ultimate–Proximate Distinction in the Human Behavioral Sciences. Perspectives on Psychological Science 6(1):38-47.

Tilly, Charles
2001    Relational Origins of Inequality. Anthropological Theory 1(3):355-372.

2008    Explaining Social Processes. Paradigm Publishers, Boulder, CO.

Tilly, Charles and Robert E. Goodin
2006    It Depends. In Oxford Handbook of Contextual Political Analysis, edited by Robert E. Goodin and Charles Tilly, pp. 3-32. Oxford University Press, New York.


Marcus said...

A couple of points:

1. Terms like causation and mechanism smack of the physical sciences, without considering how those sciences may have moved on since Newton. There's a lot going on in quantum physics and so on that doesn't make its way to the social sciences. So let's not pretend to be on their level.

2. Each social phenomenon is likely to be multi-causal, and not only in a proximate/ultimate cause way. This is what the Annales school and time perspectivism have taught us. This implies it's more interesting to look at causation using a systemic approach, since these can bring together the different causal factors in a coherent framework.

3. There is a bias towards propensity and mechanism based accounts, because they describe how things work within a system. Whereas systemic explanations seem much more tedious and of unclear value. But if one accepts the idea of multiple, interacting causes it is impossible to speak of a closed system. The only way propensity and mechanis accounts can work, in my view, is when quite artificial boundaries are set up that specify precisely what has to be accounted for.

So, while I think it's possible to do something with causation, I also believe it's near value-less if not put within a wider systemic framework. Actually, I still think Childe has the best ideas so far about this in archaeology. Doesn't say much good about epistemology in our profession, probably. Though of course the empirical work is excellent today.

Michael E. Smith said...

@Marcus- I would suggest reading some of the social science literature on causal mechanisms before dismissing the concept for archaeology.

Marcus said...

I did browse through it at one point, primarily through the Understanding Society blog. It didn't entice me to look further. The thrust seems to be to replace the larger social categories (early civilisations, capitalism etc.) with a focus on social mechanisms. In my view that impoverishes analysis if one takes it as the central core of one's method.

Not saying it's useless, especially in modern contexts, I'm just more of a systemic guy.

Neuroskeptic said...

"Terms like causation and mechanism smack of the physical sciences"

But where did the physical sciences got those words from initially? From a pre-scientific understanding of everyday life... so they seem like a good place to start, when you're trying to understand the everyday life of the past.

Marcus said...

Neuroskeptic, I'm definitely not against archaeology learning anything from the physical sciences. But their ideas about causality are a little more developed than in the social sciences, I would say. We could learn from that, actually, if we got through the mathematics.