Negated concepts interfere with anaphor resolution

Abstract

Across three experiments, we studied whether the mental representations of negated concepts are suppressed. In two reading-time experiments, we tested whether the presence of a negated nonreferent distractor (e.g., Justin bought a mango but not an apple. He ate the fruit.) interfered with the process of anaphor resolution. We found evidence that highly-typical category exemplars (e.g., apple) in the negated nonreferent role interfere with anaphor comprehension; evidence regarding less-typical category exemplars was mixed. In a third experiment, participants read brief passages like those from the prior experiments and their memory for the category exemplars was tested in a surprise cued-recall task. Once again, we found evidence that negated nonreferents are considered during anaphor resolution. These results are inconsistent with a theoretical perspective that posits that negation of a concept obligatorily leads to suppression of that concept. Instead, we argue that the comprehension of negation will be dictated by its pragmatic role.

Activation of negated and non-negated entities

“…According to the pragmatic inference hypothesis (Levine and Hagaman, 2008), negated concepts are less difficult to process when they have been previously mentioned because the need to infer a presupposition to be canceled or denied is made easier or removed entirely. That is, given the sentence, Jones does not own a Porsche, we contend that the time course of the processing of an isolated negation like this starts with what is asserted by the sentence, and then after that continues with how the sentence fits into the larger context, to the extent a context exists.…”

“…The non-licensing context was two sentences long and provided some background for the target sentence, although there was no obvious relation between the context and specific items in the target sentence. The target sentence contained two direct objects from the same taxonomic category, one of which was a high-frequency exemplar (e.g., apple) and the other was a low-frequency exemplar (e.g., mango); typicality has been shown to play a role in the activation level of negated entities (Levine and Hagaman, 2008). The target sentence appeared in one of two conditions.…”

“…Other results (e.g., Autry and Levine, submitted for publication; Giora et al, 2007;Kaup, 2001;Kaup and Zwaan, 2003;Levine and Hagaman, 2008;Shuval and Hemforth, 2008) suggest a more-complex psychological theory of representation of negation is needed, but for at least the representation of a simple sentence with a negated predicate in which the absence of an entity is asserted, DRT provides an excellent starting point.…”

“…Critical to the argument here is that subjects had a tendency to provide reasons why there were not more football fans at the game; that is, they provided a proposition to be denied by the negative operator. Levine and Hagaman (2008) labeled this tendency to generate presupposed information in the face of unmotivated negation the pragmatic inference hypothesis (see also Tian et al, 2010). Levine and Hagaman examined activation levels of negated concepts in materials similar to those used by Kaup (2001) and MacDonald and Just (1989) (see (2) above).…”

“…Based on the pragmatic inference hypothesis (Levine and Hagaman, 2008), it was predicted that Noun1 (i.e., the first direct object) would be written about equally often in the NOT and AND conditions, since it is non-negated in both, and that Noun2 (i.e., the second direct object) would be written about more often in the NOT condition, when it is negated, than in the baseline AND condition, when it is non-negated. In the sample passage given in Table 1, there is no reason to expect Justin to buy an apple, so it appears infelicitous for the sentence to read but not an apple.…”