The last 10 years (well… really, it’s just the last 7, but I’m assuming I’ll spend a few more years on it also!).

The ultimate aim of using a language is to communicate the world around us. But that world is constantly changing, as are the objects it contains. Much of our linguistic effort goes into describing the changes that do, or could, happen in that world. That is, we describe events in which someone or something acts upon someone or something else, over some period of time, and often with one or more consequences. These more recent years of my research have been spent considering the consequences of having to keep track of those … er …. consequences, and specifically, of having to keep track of and represent the changes that objects undergo as events unfold. To put it in really concrete terms: Consider the following example
  • The squirrel will crack the acorn, but first it will lick the acorn
  • The squirrel will crack the acorn, and then it will lick the acorn
Here, one has to represent both states of the acorn - the original uncracked state, and the subsequent cracked state. And depending on whether the second part of the sentence starts ‘but first’ or ‘and then’, the licking is done to one or other ‘version’ of the acorn. So the question I’m interested is, and have been exploring using a variety of different techniques (eye-tracking and fMRI primarily) concerns the costs associated with having to maintain two (or more) mutually exclusive representations of the same object in distinct states. They’re mutually exclusive because, in the above example, when selecting which acorn is licked, it’s one or the other of the two versions. It can’t be both. It turns out that having to represent and/or select from between these different versions causes those parts of the brain to ‘light up’ that also light up when there’s other kinds of conflict (e.g. words with multiple meanings that you have to decide between). These bits of the brain don’t light up simply because there’s two things (in this case, two versions of the acorn) rather than one.

So that was interesting enough - the finding that representing the ‘before’ and ‘after’ of an object engenders conflict. But even more interestingly, we found that the more dissimilar the distinct states of the same object, the more conflict we observed in these ‘conflict areas’ of the brain (specifically, a part of frontal cortex called the left inferior frontal gyrus… if you’re a brain person, that will mean something to you given all the other stuff that’s known about this brain area). This is interesting because when one has to represent/select distinct objects, the more similar they are the greater the conflict. So whereas for distinct objects one typically observes similarity-based interference, for distinct states of the same object, we observe dissimilarity-based interference.

More recently, this work has evolved into a theory of how we represent events and the objects that take part in those events. More specifically, it focuses now on the interplay between knowledge of types of objects (e.g. acorns as a class of things) and knowledge of specific individual instances of those things (aka "tokens" - e.g. the specific acorn that was cracked, as distinct from some other that was not). This work is described in the 3rd paper referenced below (from 2017).

This work would not have been possible without my collaborators Sharon Thompson-Schill, at University of Pennsylvania, and Nick Hindy, now at Princeton. You can read about it here:

References

Hindy, N.C., Altmann, G.T.M., Kalenik, E., & Thomspon-Schill, S.L. (2012). The effect of object-state changes on event processing: Do objects compete with themselves? The Journal of Neuroscience, 32(17), 5795–5803.

Altmann, G.T.M. (2013). Anticipating the garden path: The horse raced past the barn ate the cake. In M. Sanz, I. Laka, M. Tanenhaus (Eds.) Language Down The Garden Path: The Cognitive and Biological Basis for Linguistic Structure. Oxford: Oxford University Press.

Altmann, G.T.M. (2017). Abstraction and generalization in statistical learning: implications for the relationship between semantic types and episodic tokens. Philosophical Transactions of the Royal Society B 372: 20160060. http://dx.doi.org/10.1098/rstb.2016.0060