Elsevier

NeuroImage

Volume 43, Issue 2, 1 November 2008, Pages 358-367
NeuroImage

Negation in the brain: Modulating action representations

https://doi.org/10.1016/j.neuroimage.2008.08.004Get rights and content

Abstract

Sentential negation is a universal syntactic feature of human languages that reverses the truth value expressed by a sentence. An intriguing question concerns what brain mechanisms underlie our ability to represent and understand the meaning of negative sentences. We approach this issue by investigating action-related language processing and the associated neural representations. Using functional magnetic resonance imaging we measured brain activity in 18 healthy subjects during passive listening of sentences characterized by a factorial combination of polarity (affirmative vs. negative) and concreteness (action-related vs. abstract). Negation deactivated cortical areas and the left pallidum. Compared to abstract sentences, action-related sentences activated the left-hemispheric action-representation system. Crucially, the polarity by concreteness interactions showed that the activity within the action-representation system was specifically reduced for negative action-related vs. affirmative action-related sentences (compared to abstract sentences). Accordingly, functional integration within this system as measured by Dynamic Causal Modeling was specifically weaker for negative action-related than for affirmative action-related sentences. This modulation of action representations indicates that sentential negation transiently reduces the access to mental representations of the negated information.

Introduction

Sentential negation is a universal syntactic feature of human languages that mediates the expression of the reverse truth value of any given sentence. Across languages it is invariably grammaticalized by using lexical morphemes, such as “not” in “John has not left” vs. “John has left” (Horn, 1989, Zanuttini, 1997). In other words, in no language can a negative sentence be realized by simply rearranging the word order of the corresponding affirmative declarative sentence. This sharply contrasts with cases like interrogative sentences, such as “Has John left?” derived from its affirmative counterpart “John has left”. Abundant psycholinguistic research has been devoted in the past to investigate how sentential negation affects language processing. Two main classes of results have emerged from these studies. On the one hand, sentence comprehension – as tested for example in sentence-picture matching tasks (Carpenter and Just, 1975, Clark and Chase, 1972, Trabasso et al., 1971) – was shown to be more difficult for negative than for affirmative sentences. Negative sentences required longer processing times and were associated with higher error rates. On the other hand, studies testing for the impact of negation on the accessibility of information mentioned within its scope – using for instance word recognition or priming tasks (MacDonald and Just, 1989, Kaup, 2001, Kaup and Zwaan, 2003) – showed that response times were significantly longer for negated than for non-negated items. These two classes of evidence have led to divergent views on the mental representation of negation. The first type of evidence has been taken to suggest that negative sentences are syntactically more demanding than affirmative sentences, and according to this view stronger activations of perisylvian brain regions are to be expected. In turn, the second type of evidence speaks for a reduced access to conceptual representations of the negated items; accordingly, reduced activations of the brain circuits involved in conceptual representations are to be expected.

The neural correlates underlying the processing of syntactic negation are still largely unknown. Two different fMRI studies have tested the hypothesis of a higher computational load for negative compared to affirmative sentences. In one study, negative vs. affirmative sentences describing visuo-spatial relations were compared. Significantly higher activations in the left posterior temporal and bilateral posterior parietal lobes were found for negative sentences (Carpenter et al., 1999). In the second study on bilingual subjects, higher signal for negation vs. affirmation was found in left perisylvian regions, but only in the participants' second language, which was mastered at a moderate level of proficiency, and not in their first language (Hasegawa et al., 2002). Altogether, it seems that negation elicited higher activations only in combination with increased extrinsic, non-linguistic task demands, such as the processing of visuo-spatial relations or of a second language at low proficiency.

An important unresolved question regards whether the impact of negation on the neural responses elicited by the negated propositions is dependent or independent from the semantic field involved. For example – based on the evidence of partially segregated anatomo-functional correlates for concrete vs. abstract semantic contents (Binder et al., 2005, Kiehl et al., 1999, Martin-Loeches et al., 2001, Perani et al., 1999, Sabsevitz et al., 2005, Tyler et al., 2001) – if negation determines a reduced access to the mental representations of negated concepts, anatomically distinct modulatory effects for concrete or abstract semantic content should be expected. The present experiment addresses precisely this issue, hinging on language simulation theories, which postulate that language comprehension is mediated by implicit sensorimotor simulations of the content described by linguistic utterances (Barsalou, 1999, Glenberg and Kaschak, 2002). Mental simulation mechanisms have been specifically implicated with respect to the role of the action-representation system in embodied language representations (Gallese, 2007), with mirror neurons providing a plausible neurophysiological substrate (Rizzolatti and Craighero, 2004). A growing number of studies has consistently shown that linguistic utterances describing actions performed by different body parts activate the same action-representation circuits which subserve the execution and the observation of the actions described (Pulvermuller, 2005). Somatotopically organized effects in the left premotor cortex have been found with fMRI for mouth-, hand-, and leg-related verbs (Hauk et al., 2004), phrases (Aziz-Zadeh et al., 2006), and sentences (Tettamanti et al., 2005). In the latter study, activations for action-related sentences compared to abstract sentences were not confined to the left premotor cortex, but extended into the entire left fronto-parieto-temporal system subserving action representation, including the inferior parietal lobule and the posterior temporal cortex. In turn, abstract sentences compared to action-related sentences were specifically associated with an effect in the posterior cingulate cortex.

If negation indeed determines a reduced access to the specific semantic information contained in the predicate of the negated propositions, we should expect two distinct effects: 1) a reduced haemodynamic response in the left fronto-parieto-temporal system representing actions for negative vs. affirmative action-related sentences; and 2) a reduced response in the posterior cingulate cortex for negative vs. affirmative abstract sentences. As a consequence of the content-specific activation reduction, we should also expect that the functional integration between the brain regions constituting the left fronto-parieto-temporal system as measured by effective connectivity (Lee et al., 2006) are reduced in the context of negative action-related sentences. In other words, we should expect a stronger, synergistic increase of activity in the brain regions constituting the action-representation system for affirmative vs. negative action-related sentences. Conversely, if negation is associated with higher syntactic computational loads, we should expect stronger, content-independent regional haemodynamic responses for negative vs. affirmative sentences, possibly in left perisylvian areas.

In an event-related fMRI experiment, 18 participants passively listened to declarative sentences (Supplementary Table 1) characterized by the factorial combination of polarity (affirmation or negation) and concreteness (action-related or abstract). Based on this factorial design, we addressed two independent issues: the anatomo-functional correlates of sentential negation, and the interaction between polarity and concreteness. We assessed both functional specialization, under classical General Linear Model assumptions, and functional integration, using Dynamic Causal Modeling (DCM) (Friston et al., 2003). Abstract sentences described operations in which no physical entities were involved, such as “Ora apprezzo la fedeltà” (“Now I appreciate loyalty”). Action-related sentences, instead, described actions involving physical entities, such as “Ora premo il bottone” (“Now I push the button”).

Section snippets

Subjects

Eighteen right-handed volunteer subjects (12 females, mean age 24.7 years, range 20–34 years) of comparable education level (Graduate Level) took part in the experiment. Participants were all native monolingual speakers of Italian, with no history of neurological or psychiatric disorders and no structural brain abnormalities. They gave written consent to participate in the study after receiving an explanation of the procedures. The study was approved by the Ethics Committee of the San Raffaele

Behavioral data collected after fMRI data acquisition

Once the fMRI data acquisition was completed, all participants declared being unaware of the grouping of sentences into the action-related vs abstract condition. Some of them, however, noticed that the same sentences were presented both in the affirmative and in the negative form. Immediately after fMRI data acquisition, participants were also asked to recall the highest number of sentences or part of sentences they could remember. On average they were able to recall 9.24% (SD = 4.90) of all the

Summary

In summary, irrespective of the level of concreteness (action-related vs. abstract), sentential negation was associated with a deactivation of pallido-cortical areas. Crucially, negation induced distinct effects for action-related vs. abstract sentences. In the case of negative action-related sentences, a reduction of both activations and connection strengths occurred within a left-hemispheric fronto-parieto-temporal system. For negative abstract sentences, there was a deactivation of the

Discussion

The present results represent a preliminary step toward the understanding of the neural mechanisms of sentential negation. Overall, the results do not support the hypothesis of a greater processing load associated to negative sentences, and are in line with the idea of a reduced access to the negated information. A theoretical model on negation has been recently advanced (Kaup et al., 2007), which assumes that the process of understanding a negative sentence (e.g. “John has not left”) can be

Acknowledgments

We thank Hellen M. Della Justina and Maria Giavazzi for help with the DCM analysis, Jubin Abutalebi, Giacomo Rizzolatti, and Cristina Saccuman for helpful comments on this manuscript. Supported by the Italian Ministry of University and Research (FIRB 2003119330-009) and by the European Community (“ABSTRACT” — STREP FP6-2004-NEST-PATH-HUM).

References (50)

  • LeeL. et al.

    Large-scale neural models and dynamic causal modelling

    Neuroimage

    (2006)
  • MaceyP.M. et al.

    A method for removal of global effects from fMRI time series

    Neuroimage

    (2004)
  • Martin-LoechesM. et al.

    Functional differences in the semantic processing of concrete and abstract words

    Neuropsychologia

    (2001)
  • PennyW.D. et al.

    Comparing dynamic causal models

    Neuroimage

    (2004)
  • SabsevitzD.S. et al.

    Modulation of the semantic system by word imageability

    Neuroimage

    (2005)
  • TrabassoT. et al.

    Storage and verification stages in processing concepts

    Cogn. Psychol.

    (1971)
  • VitaliP. et al.

    Generating animal and tool names: an fMRI study of effective connectivity

    Brain Lang.

    (2005)
  • BarsalouL.W.

    Perceptual symbol systems

    Behav. Brain Sci.

    (1999)
  • BenjaminiY. et al.

    Controlling the false discovery rate: a practical and powerful approach to multiple testing

    J. R. Stat. Soc. B (Meth.)

    (1995)
  • BinderJ.R. et al.

    Distinct brain systems for processing concrete and abstract concepts

    J. Cogn. Neurosci.

    (2005)
  • CarpenterP.A. et al.

    Sentence comprehension: a psycholinguistic processing model of verification

    Psychol. Rev.

    (1975)
  • CoplandD.

    The basal ganglia and semantic engagement: potential insights from semantic priming in individuals with subcortical vascular lesions, Parkinson's disease, and cortical lesions

    J. Int. Neuropsychol. Soc.

    (2003)
  • DaleA.M.

    Optimal experimental design for event-related fMRI

    Hum. Brain Mapp.

    (1999)
  • De MauroT. et al.

    Lessico di frequenza dell’ italiano parlato

    (1993)
  • GalleseV.

    Before and below ‘theory of mind’: embodied simulation and the neural correlates of social cognition

    Philos. Trans. R. Soc. Lond. B Biol. Sci.

    (2007)
  • Cited by (0)

    View full text