When Pigs Fly But Hell Hasn't Frozen Over: Semantic Anomalies, Context, and the Inferior Frontal Gyri


When reading the title of this post, your knowledge of the world was sufficient to let you interpret the phrase "when pigs fly," but also alerted you to the fact that it is inconsistent with much of that world knowledge: clearly, pigs don't fly. A new study by Menenti, Petersson, Scheeringa & Hagoort localizes the neural basis of this "anomaly processing" to a particular region of the prefrontal cortex - the inferior frontal gyri - and finds that the local context surrounding such anomalies (say, that you're told hell has frozen over) has much stronger effects in one hemisphere than the other.

To assess anomaly processing, 32 subjects in an fMRI scanner saw critical sentences like the following: "With the lights on, you can see [more/less] at night." In this example, "less" is a semantic anomaly (typically, lights help you see more), unless it occurs in the context of (for example) a discussion of the problem of light pollution in astronomy. Pitting world knowledge against local context, Menenti et al were able dissociate the neural activity in response to semantic anomalies from the activity in response to expected phrases, and could further dissociate how those differences were modulated by local context. To ensure that subjects were paying attention, 10% of the trials involved a "pop quiz" about the previous sentence.

A fairly restricted set of brain regions responded to these manipulations - in particular, the left and right inferior frontal gyri, regions very similar to those commonly associated with Broca's aphasia (a disorder in which patients appear to select words only with great effort, and may be incapable of using or interpreting complex grammar). See this video for a sad but classic example of the importance of this region of the brain:

In the Menenti et al study, both the left and right inferior frontal gyri showed greater responses to anomalies than expected words in neutral contexts, and this response was reduced when the local context made the anomalies more acceptable. However, the reduction was far greater in the right hemisphere (in fact, only the lowest portion of the left side even showed a trend in this direction), suggesting that the right inferior frontal gyrus may have a fairly special role in processing anomalies with respect to the local context.

Of course, it's hard to know exactly what these regions are doing, since the subjects weren't given any explicit task. On the bright side, we can always speculate. Here are a few possibilities:

1) Perhaps both the inferior frontal gyri are concerned with selecting information in the presence of competition (consistent with dominant accounts of the left IFG), such that it takes more neural activity to access the meaning of semantically anomalous words. Processing difficulty in the left IFG is best described with something like direct word-to-word association strength (how can I use the last word to help me access the meaning of the current one), whereas difficulty in the right side is better described with a more temporally-protracted measure of association (how can I use the last 50 words to help me access the meaning of the current one). Thus, context cues reduce demand on the right without influencing them too much on the left.

2) A related but alternative explanation is that the left inferior frontal gyrus is responsible for selecting information with semantic value, and the right side is additionally recruited only when that selection process is particularly difficult. In the case of Menenti et al, the right inferior frontal gyrus may be most strongly recruited when to-be-accessed semantic information is highly anomalous. Thus, in contexts where local constraints on semantic information make the job of selection easier, right IFG's role is greatly diminished.

The authors endorse an explanation based on the greater involvement of right hemisphere regions in the processing of novel or inferred information and in the processing of longer temporal episodes.

More broadly, this kind of "anomaly processing" is associated with at least two sharp changes in voltage on the surface of the scalp, which distinguish between semantic anomalies ("i eat the shoe"; see also) and grammatical anomalies ("i eating the pie") or musical discordance. The latter is sometimes considered a delayed version of the "oddball effect," in which any contextually-infrequent stimulus is associated with a particular positive deflection in scalp-recorded electrical potentials. It remains to be seen how these electrical potentials relate to the kinds of neural activity observed by Menenti et al. with fMRI.

Laura Menenti, Karl Magnus Petersson, René Scheeringa, Peter Hagoort (2008). When Elephants Fly: Differential Sensitivity of Right and Left Inferior Frontal Gyri to Discourse and World Knowledge Journal of Cognitive Neuroscience, Early Access (Early Access), 2147483647-11 DOI: 10.1162/jocn.2008.21163

Addendum: A third, more radical possibility for the Menenti results is that the left and right inferior frontal gyri implement completely distinct functions (bear with me, Ockhamites). For example, the left side may be involved in selection, and the right side in indicating when those selection criteria must change. That is, the right IFG could signal the presence of contextual novelty, in order to modulate the influence of left-sided regions in selecting the appropriate meaning. One such modulation could be the application of a general "brake" to neuronal processing, allowing lateral competition among representations to settle on a single interpretation. I note that this mechanism might itself underlie the explanations above, although the converse is not true (so there, Ockhamites!).

More like this

A lot has been written about domain-general processing in prefrontal cortex, and a very old lesson often gets overlooked: there are very basic hemispheric asymmetries (particularly in PFC) that divide information processing by modality. A very nice study by Morimoto et al provide a nice reminder…
How does memory help to accomplish moment-to-moment goal-directed action? Classic accounts, such as Baddeley's working memory model, suggest that there are separate storage and processing ("executive") mechanisms, whereas newer accounts (proposed by a variety of researchers) propose that storage…
"Instead of trying to produce a programme to simulate the adult mind, why not rather try to produce one which simulates the child's?" - Alan Turing (Computing Machinery, p456) One of the defining features of childhood cognition is "behaving without thinking." Not surprisingly, developmental…
Something's afoot when a massively parallel and distributed system shows a bottleneck in performance. We've known that numerous bottlenecks plague cognition since the 1940's, but only with recent advances in neuroimaging have we been able to say whether these bottlenecks reflect the intrusion of…