Talk:N400

This is (for the most part) an excellent review of the N400 literature for the novice. The two authors are eminent experts in Event-related potentials in language processing and give a fair and unbiased overview over the different accounts put forward regarding the N400.

I have just one request: I do not think that the detailed history at the beginning of the article serves its purpose. First of all, the discussion of the positive component distracts from the main message and is confusing. I also do not think that the different initial (and partially inappropriate) interpretation need to be reiterated here. If the authors want to keep the history section, it should be shortened and maybe moved to another place. The reader should get the recent and better substantiated knowledge first.

Also, the authors may want to discuss whether on can speak about "the" N400 in light of the different scalp distribution in more detail.

But as I said before, this is a very good description of the N400.

AUTHORS" RESPONSE:

We have shortened the history section rather than removing it altogether. We thought it important to keep the bit about the P3b, because many who know almost nothing about ERPs do tend to have heard of the P3b. They are the ones who ask what’s the difference between the N400 and the P3b? Isn’t the P3b also about surprise? We have made the P3b and N400 contrast more explicit by noting that unexpected physical items (large font words) elicit P3b activity and that both the N400 and P3b may be recorded to the same stimulus.

The issue of whether there is “the N400” is fairly complex, and we were already too long given article specs. We feel that a discussion of that issue is beyond the scope this article. We are addressing it in a forthcoming article, and can add that as a reference in this Scholarpedia piece once we are done.

Reviewer 2:

This is an excellent overview of the N400 event-related potential written by two authors who have made seminal contributions to our understanding of its functional relevance. Below are a few suggestions:

Early on, it may be worth defining the ‘N400 effect’ as the modulation of N400 amplitude between two conditions, as this is sometimes a source of confusion to people who are not familiar with this literature.

The authors talk about ‘sentence level expectancy manipulations’ followed by ‘priming paradigms’ in an initial overall section called ‘Main Paradigms’. I would shorten these sections and confine them to describing the structure of the experimental paradigms themselves. Some of what is covered could be moved to later sections that are covered under ‘Factors that influence N400 amplitude’ to avoid redundancy. For example, the discussion of whether the N400 is sensitive to level of expectancy in plausible sentences could be moved to the section on ‘expectancy/cloze probability’. Indeed, it may be worth subdividing this section into one covering sentences and another covering discourse. Also, in this section, the authors could mention the fact that an N400 effect is evoked to words in sentences describing situations that are incongruous with ‘real world’ knowledge e.g. “Every morning at breakfast the boys would bury…” (vs. eat), Kuperberg et al. 2003 Cognitive Brain Research; “The Dutch trains are white (vs. yellow), Hagoort et al. 2003 Science. This, I realize, falls into the general idea that less expected words evoke an N400, but much has been made of the sensitivity of the N400 to our knowledge of events or truths in the real world, and so it may be worth explicitly highlighting this.

The section ‘Factors that do not influence N400 amplitude’ is tricky. I certainly think that it is important to emphasize the fact that the amplitude of the N400 does not necessarily reflect degree of contextual constraint. I’m not sure that the other sections should fall here, however. The work by Staab et al. and by Nieuwland and Kuperberg make it clear that the N400 amplitude often does reflect the semantic “truth value” of a given expression, so long as that expression is pragmatically licensed, so implying that the N400 is not sensitive to negation/quantification and starting out with the older studies “A robin is a tree” may be a bit misleading. In fact, I’d include pragmatic licensing as a factor that the N400 is sensitive to in the previous section and refer to the studies by Staab et al. and by Nieuwland and Kuperberg, as well as studies by van Berkum (together with his recent book chapter.)

Regarding the N400 and ‘thematic role violations’, this is again tricky. It is certainly true that certain violations of expected thematic role do not necessarily evoke an N400 effect but the story is complex and robust N400 effects are certainly seen to such violations in some situations. I think that I’d abstract a bit away from ‘thematic role violations’ themselves and talk more generally about ‘plausibility’ as a construct. In other words, while the amplitude of the N400 is often sensitive to ‘expectancy’, it doesn’t not necessarily reflect degree of plausibility with respect to real-world knowledge. There are some situations in which highly implausible words evoke an N400 that is smaller than that evoked by semi-implausible words (e.g. “At breakfast the eggs would eat..” vs. “At breakfast the boys would eat…”), Kuperberg et al. Cognitive Brain Research 2003, reviewed by Kuperberg 2007 Brain Research; there are other situations where the amplitude of the N400 is equal between very implausible and semi-implausible words (e.g. Van de Meerendonk et al. JCN). There are other situations where the N400 is the same amplitude to plausible sentences and very implausible sentences (e.g. “The author finished the book…” versus “The author astonished the book…” in a recent paper on complement coercion by Kuperberg et al. JCN in press.)

Under ‘neural sources’ of the N400, the authors might consider mentioning some of the fMRI studies that have used ‘N400-like’ paradigms e.g. Kuperberg et al., 2003 JCN; Hagoort et al. Science 2003, implicating both temporal and left inferior cortices as contributors to this component (although here it would also be important to acknowledge the caveats in interpretation given the slow time-course of BOLD activity). Lau et al. 2008 Nature Reviews Neuroscience provide a review of some of these studies.

Small point:

The N400 and movies: may be worth also referencing the latest study by Sitnikova et al. which better addresses the functional implications of the N400 to movie stims than the original 2003 short paper: Sitnikova T, Holcomb PJ, Kiyonaga KA, Kuperberg GR. Two neurocognitive mechanisms of semantic integration during the comprehension of visual real-world events. J Cogn Neurosci Nov 2008;20(11):2037-2057.

AUTHORS' RESPONSE

We have reorganized the paradigms section per request. We shortened the text and removed the section headings, so that now the paradigms are described on a continuum from single stimulus to discourse.

We have added explicit reference to the sensitivity of the N400 to world knowledge.

We slightly modified (see final sentence) the section on negation. As an aside, we don’t actually think that it is misleading to start with the older studies. It could have been the case that the N400 was sensitive to true/false or to plausibility in all cases. It clearly is not, and it’s important for those who want to understand and/or use the N400 to know this (we now emphasize that this is the case under simple sentence verification paradigms). The fact that under other circumstances the N400 is sensitive to true/false is also important in part by contrast, and is something we originally said as well and that we now emphasize more.

The section on N400 and thematic violations also was reworked, and, as suggested, framed with respect to plausibility. We now explicitly state that the thematic role case is an example of a number of cases in which N400s do not pattern with plausibility, and that this pattern comes out in different ways across the literature. This subset of the literature is especially complicated, and clearly we could not do it all the justice it deserves, but we’ve tried.

We did not change the section on the neural sources, and we feel it is beyond the scope of this overview to talk about fMRI. The mapping between the two measures is complex and we feel that this is not the place to go into that complexity.

We have updated the reference to Sitnikova et al.’s work.