The interplay between mood and language comprehension: Evidence from P600 to semantic reversal anomalies

Neuropsychologia 51 (2013) 1027–1039

Contents lists available at SciVerse ScienceDirect

Neuropsychologia journal homepage: www.elsevier.com/locate/neuropsychologia

The interplay between mood and language comprehension: Evidence from P600 to semantic reversal anomalies Constance Th.W.M. Vissers a,b,n, Uli G. Chwilla b, Jos I.M. Egger a,b,c, Dorothee J. Chwilla b a

Centre of Excellence for Neuropsychiatry, Vincent van Gogh Institute for Psychiatry, Venray, The Netherlands Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen, The Netherlands c Behavioural Science Institute, Radboud University Nijmegen, The Netherlands b

a r t i c l e i n f o

abstract

Article history: Received 30 July 2011 Received in revised form 10 December 2012 Accepted 17 February 2013 Available online 27 February 2013

Little is known about the relationship between language and emotion. Vissers et al. (2010) investigated the effects of mood on the processing of syntactic violations, as indexed by P600. An interaction was observed between mood and syntactic correctness for which three explanations were offered: one in terms of syntactic processing, one in terms of heuristic processing, and one in terms of more general factors like attention and/or motivation. In this experiment, we further determined the locus of the effects of emotional state on language comprehension by investigating the effects of mood on the processing of semantic reversal anomalies (e.g., ‘‘the cat that fled from the mice’’), in which heuristics play a key role. The main findings were as follows. The mood induction was effective: participants were happier after watching happy film clips and sadder after watching sad film clips compared to baseline. For P600, a mood by semantic plausibility interaction was obtained reflecting a broadly distributed P600 effect for the happy mood vs. absence of a P600 for the sad mood condition. Correlation analyses confirmed that changes in P600 in happy mood were accompanied by changes in emotional state. Given that semantic reversal anomalies are syntactically unambiguous, the P600 modulation by mood cannot be explained by syntactic factors. The semantic plausibility by mood interaction can be accounted for in terms of (1) heuristic processing (stronger reliance on a good enough representation of the input in happy mood than sad mood), and/or (2) more general factors like attention (e.g., more attention to semantic reversals in happy mood than sad mood). & 2013 Elsevier Ltd. All rights reserved.

Keywords: Mood Heuristic processing Semantic reversal anomalies P600 effect

1. Introduction Research from different domains of psychology like perception, memory and decision making has shown that our background emotional state influences the way in which we process information (Clore & Huntsinger, 2007). A positive mood is characterized by the ability to flexibly take alternative cognitive perspectives (Ashby, Turken, & Isen, 1999). Positive mood validates accessible cognition and leads to a more global, category level of processing of information, that is, relating incoming information to what is already known based on our world knowledge including schemas and stereotypes (e.g., Gasper & Clore, 2002; Kimchi & Palmer, 1982). Negative mood, on the other hand, invalidates accessible cognition and is characterized by local, item specific processing, with close attention to the details (e.g., Schwarz, 2002). In line with the inclination for global processing in a happy mood and local processing in sad mood, it has been shown that happy people are more likely than sad

n Corresponding author at: Centre of Excellence for Neuropsychiatry, Vincent van Gogh Institute for Psychiatry, Stationsweg 46, 5803 AC Venray, The Netherlands. Tel.: þ 31 478 527 552, mobile: þ 31 641 227 990. E-mail address: [email protected] (C.Th.W.M. Vissers).

0028-3932/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.neuropsychologia.2013.02.007

people to rely on heuristics when processing information. A heuristic is a mental shortcut that allows people to solve problems and make judgments quickly and efficiently (see also below). Positive mood has been shown to favor the use of heuristics (see for a review: Clore & Huntsinger, 2007). For instance, in a situation that activates a schema, people in a happy mood are more likely than those in a sad mood, to use the schema and rely on their general knowledge structures to fill in the blanks, when recalling the details of the situation (Bless et al., 1996). Furthermore, it has been demonstrated that sad participants are more likely than happy participants to utilize knowledge about others to make inferences about their mental states on Theory of Mind tasks. The differential effect of mood on Theory of Mind performance suggests that happy people tend to rely on an egocentric default whereas sad people tend to incorporate knowledge about others more deliberately (Converse, Lin, Keysar, & Epley, 2008). While research on the interface of affect and cognition has generally established that sad people are more inclined to use a systematic strategy and happy people are inclined to rely on heuristics (see Bless & Schwarz, 1999; Gasper & Clore, 2002; Kimchi & Palmer, 1982; Schwarz, 2002 for a review) little yet is known about the interplay between mood and language processing.

1028

C.Th.W.M. Vissers et al. / Neuropsychologia 51 (2013) 1027–1039

In two recent Event-Related Potential (ERP) experiments we have demonstrated interactions between mood and semantic processing (Chwilla, Virgillito, & Vissers, 2011), and between mood and the processing of syntactic verb-agreement violations (Vissers et al., 2010). The goal of this article is to further elucidate which processes contribute to the effects of emotional state on processes of language comprehension. To this aim, we explored the effects of a participant’s background emotional state on the processing of semantic reversal anomalies in which, as we will outline below, heuristic processing plays a key role. 1.1. Heuristic processing in language comprehension Language comprehension is a highly complex process in which information from several sources like pragmatic, semantic and syntactic information have to be represented, retrieved and combined to understand a sentence. Despite this and the frequent occurrence of ambiguities at several levels like the syntactic and semantic level, we rarely experience difficulties in understanding sentences in our mother tongue. People comprehend sentences rapidly and without conscious effort. Theories about the way in which this linguistic information is represented, retrieved and combined during language processing have urged different lines of psycholinguistic research during the last decade. In the 1960s, most psycholinguists shared Chomsky’s view that syntactic processing proceeds independently of semantic information. Fodor (1983) assimilated the concept of independent syntactic processing into the general idea that syntactic processing is modular. Modular processes are algorithmic, autonomous and operate bottom-up. These processes are not guided by world knowledge, beliefs or expectations. So, according to Fodor, syntactic processes are not guided by the meaning or plausibility of the sentence. The so called syntax-first theories embody the abovementioned claim that the syntactic module is restricted to the domain of syntactic information and is immune to non-syntactic information. Constraint satisfaction models have been proposed as a competitor of syntax-first models. According to these models, information from all sources interacts continuously during the formation of an internal interpretation. That is, all kinds of information provided by the incoming words (e.g., context, discourse and semantic information) can jointly affect the activation of different syntactic alternatives (e.g., McDonald, Maryellen, Pearlmutter, & Seidenberg, 1994; St. John & McClelland, 1990; Trueswell & Tanenhaus, 1994). Ever since Fodor (1983) argued that parsing is basically a reflex, most models of sentence comprehension have assumed that an interpretation must be based on an initial syntactic structure, even if the activation level of that structure can be influenced by nonsyntactic sources of information. This implies that, according to both syntax-first models and constraint satisfaction models, every interpretation is complete, detailed and accurate and cannot be based on shallow, inaccurate or incomplete processing. However, the meaning people derive from a sentence is often not a reflection of its true content. Researchers in other domains of cognition, such as decision making and reasoning, have argued that human behavior is at least partially driven by heuristic processing (Gigerenzer & Goldstein, 1996; Gigerenzer, Todd, & ABC Research Group, 1999). Gigerenzer and his colleagues have proposed the idea of bounded rationality. They argue that humans often find themselves in situations which force them to make inferences about the world under limited time, knowledge, and computational power. Models of rational inference do not take these limitations into account and are therefore unrealistic. They show that simple heuristics can match or even outperform classical models of rational inference. These simple heuristics are frugal because they exploit the structure of environments and only require the use of a small proportion of the available information. As a consequence, they

are thought to be fast because information search is less computationally demanding. More recently, it has been claimed that heuristics also play a role in language comprehension. Heuristic processing implies that language users do not always take into account all relevant information, in particular, both syntactic and semantic information. In line with this, Ferreira (2003) proposes that language processing is often based on a shallow representation of the input yielding a merely ‘‘good enough’’ rather than a detailed linguistic representation of an utterance’s meaning. Therefore, Ferreira, Bailey, & Ferraro (2002), Ferreira (2003), and Ferreira and Patson, 2007 (and see also: Christianson, Luke, & Ferreira, 2010) have claimed that current models of language are missing an architectural component that can explain cases in which people exploit strategies or engage in heuristic processing of sentences that may then give rise to an inaccurate interpretation. One proposed language heuristic is referred to by Bever (1970) as ‘strategy C’, and by Ferreira et al. (2002) as ‘the plausibility strategy’. This strategy is a semantic heuristic according to which readers depend heavily on their knowledge of the meaning of individual content words, which provide a strong basis for the most plausible interpretation. Hence, according to this strategy, readers depend on schemas in long-term memory or world knowledge. Several lines of evidence support the existence of heuristic processing in language. A famous example in healthy participants is the ‘‘Moses illusion’’, reported by Erickson and Matteson (1981). When participants are asked ‘‘How many of each type of animal did Moses take on the ark?’’ most people responded ‘two’ without realizing that the proposition was false in that it was Noah and not Moses who saved the animals from the flood. Likewise, Barton and Sanford (1993) have shown that most participants do not notice an anomaly in a sentence like ‘‘Bury the survivor’’ (for a detailed discussion of incomplete more shallow processing of language meaning see Ferreira & Patson, 2007). The focus of this article is on the interplay between mood and heuristic processes in language. Consistent with the above description, the term heuristics in this article refers to the fact that language users in some situations do not take all relevant information into account, but base their interpretation on a shallow—that is, good-enough representation of the linguistic input, which is in accordance with their expectation based on wor(l)d knowledge. 1.2. The interplay between mood and syntactic processing Because the Vissers et al. (2010) study is of direct relevance for the present goals, we describe this study in more detail. In this article, we explored whether mood affects the processing of syntactic anomalies. To this aim, P600 amplitude to subject– verb agreement errors was compared to that to correct sentences (1), while ERPs were recorded. Mood was manipulated by presenting happy or sad film clips.

(1) ‘De ouders die over hun dochter spraken.’ (correct) ‘The parents who about their daughter talked [plural].’ (wordby-word translation) ‘De dochter die over haar ouders spraken.’ (incorrect) ‘The daughter who about her parents talked [plural].’ (word-byword translation) The rationale was that if emotional state influences the way in which syntactic anomalies are processed, this should be reflected by an interaction between mood and P600. The main results were

C.Th.W.M. Vissers et al. / Neuropsychologia 51 (2013) 1027–1039

as follows. First, the mood induction procedure was effective: participants were happier after watching happy film clips and sadder after watching sad film clips. Second, for P600 a mood by syntactic correctness interaction was obtained. The interaction reflected a broadly distributed P600 effect for the happy mood condition (more positive amplitudes to syntactically incorrect verbs compared to their correct counterpart) and a strong reduction in P600 effect for the sad mood condition. Correlation analyses confirmed that changes in the size of the P600 effect were accompanied by changes in emotional state. Three different scenarios were proposed that could give rise to the interaction between P600 and mood. Following the first scenario in terms of syntactic processing, mood selectively affects syntactic processes: that is, happy and sad mood could have led to increased or reduced syntactic processing, reflected in P600. According to the second scenario, mood selectively affects the use of heuristics. Relevant for that argument, it has been shown that P600 is also sensitive to heuristic factors (e.g., Coulson, King, & Kutas, 1998a, 1998b; Gunter, Stowe, & Mulder, 1997; Vissers, Chwilla, & Kolk, 2007, see below). Based on the assumption that language users have a high expectancy to read syntactically correct sentences (Vissers et al., 2007, 2010), the reduction of the P600 effect in the sad mood condition could reflect a reduced use of heuristics, whereas the increase in P600 effect for the happy mood could be due to an increased reliance on shallow processing—that is, on a good enough representation of the linguistic input. Following the third scenario, more general factors like attention or motivation could have caused the interaction between mood and P600. People in a happy mood could pay more attention/be more motivated to read sentences and, hence, detect syntactic errors, while people in a sad mood could pay less attention/be less motivated to read sentences. Future studies are clearly needed to tease apart the possible contributions of syntactic, heuristic and attentional processes that mediate the effects of a participant’s emotional state on language.

1.3. Present study The aim of the present article is to further our understanding of the relation between language and emotion. This is accomplished by investigating the effects of mood on the processing of semantic reversal anomalies in which heuristics play an important role (see below). Following Ferreira et al. (2002), Ferreira (2003), and Ferreira and Patson (2007) we assume that in language comprehension simple processing heuristics are used in addition to syntactic algorithms (e.g., Christianson et al., 2010; Ferreira et al., 2002; Ferreira & Patson, 2007; Van Herten, Chwilla, & Kolk, 2006; Vissers, Chwilla, & Kolk, 2006). An important question is how these heuristics and parsing routines are coordinated. In line with Van Herten et al. (2006), we propose that the two routes, that is, algorithmic syntactic processing and heuristics run in parallel and are largely independent from one another (see also Vissers et al., 2006). Normally heuristics and syntactic algorithms will produce the same interpretation. However, in certain sentences the two routes can clash, namely when these routes lead to conflicting interpretations. In an ERP experiment by Kolk, Chwilla, Van Herten, and Oor (2003) semantic reversal anomalies were presented which were formed by exchanging the subject and object of semantically acceptable sentences such as (2).

(2) De vos die op de stropers joeg sloop door het bos (original). The fox that at the poachers hunted [singular] stalked through the woods (literal translation).

1029

The fox that hunted [singular] the poachers stalked through the woods (paraphrase). It is clear that in these sentences the heuristic and parsing routines produce different thematic interpretations. The heuristic exploits the individual content words (fox–hunter–poachers) and comes up with a semantically plausible thematic interpretation of the sentence, that is ‘that poachers are hunting foxes’. The heuristic interpretation is highly expected based on our world knowledge. The parsing routine on the other hand proposes on the basis of a systematic analysis of the grammatical structure of the sentence that ‘foxes are hunting poachers’. Although the latter interpretation is not entirely impossible, it represents a highly unlikely event based on world knowledge. Important for the present study, these semantic reversal anomalies reliably elicit a P600 effect (Kolk et al., 2003; Vissers et al., 2006). Based on the fact that the semantic reversal anomalies are not syntactically ambiguous (both noun phrases could serve as the agent and the patient of the action expressed by the verb ending the relative clause; in the reversal sentence ‘the fox that hunted the poacher’, there is only one option: fox is the Agent, and poacher is the Theme (see also: Vissers et al., 2007) the report of a P600, which had been taken to index syntactic processing (Hagoort, Brown, & Groothusen, 1993; Osterhout & Holcomb, 1992), to these semantic reversal anomalies came as a surprise. Recently, however, several laboratories using different sentence materials and languages (English and Dutch), reported a P600 effect to semantically implausible sentences relative to their plausible counterparts (see for a review: Kuperberg, 2007, see also: Hoeks, Stowe, & Doedens, 2004; Kuperberg et al., 2003). Why would a P600 effect follow a conflict between thematic interpretations? Kolk et al. (2003) argued that the language comprehension system attempts to resolve the conflict – triggered by conflicting representations – by reanalysing the sentence to check the input for possible processing errors. In particular, the mismatch between the plausible thematic interpretation based on world knowledge and the implausible thematic interpretation based on the syntactic parse makes it necessary for the brain to re-attend the unexpected linguistic unit to check upon its veridicality. After all, an inconsistency can have two sources. It can be real, in that an unexpected event has indeed occurred (e.g. two planes crashing into the WTC buildings). On the other hand, it can also stem from a processing error. To prevent integration of erroneous information into the current discourse, the reader will generally check upon the correctness of his or her analysis in case of a conflict. This, according to the Monitoring Theory, explains the occurrence of P600 effects to semantic reversal anomalies (see for a review: Van de Meerendonk, Kolk, Chwilla, & Vissers, 2009). In this article the interaction between mood and the role of heuristics in language comprehension is investigated. To this aim, the effects of mood on the P600 effect to semantic reversal anomalies (‘the fox that hunted the poachers.’) will be compared relative to their plausible counterparts (‘the poachers that hunted the fox.’). As sketched out the use of heuristics while reading sentences with semantic reversals leads to a conflict between an expected and an unexpected sentence representation. This conflict triggers monitoring to check for processing errors and is reflected by the P600 effect. When the reader does not rely on heuristics while processing semantic reversals, no conflict and hence no monitoring response or P600 effect should be triggered. In the emotion literature, it has been shown that presentation of films is a highly effective means in inducing both positive and negative mood states (see Westermann, Spies, Stahl, & Hesse, 1996) who comparatively evaluated 11 mood induction procedures by meta-analytic procedures). Therefore, like in our previous studies (Chwilla et al., 2011; Vissers et al., 2010),

1030

C.Th.W.M. Vissers et al. / Neuropsychologia 51 (2013) 1027–1039

a participant’s mood will be manipulated by presenting film clips that either display sequential fragments from a happy movie (the animated Warner Brothers’ movie ‘‘Happy Feet’’) or from a sad movie (a drama depicting the plight of a Polish woman during the second world war, ‘‘Sophie’s Choice’’). On the assumption that we are able to successfully induce the intended changes in mood, the predictions are as follows. If mood impacts the processing of semantic reversal anomalies, this should be indicated by an interaction between mood and plausibility, as tapped by P600. As outlined above, three scenarios could underlie a mood by plausibility interaction for P600. According to one scenario, mood could modulate language processing by increasing or decreasing the contribution of syntactic processing. Based on the fact that the semantic reversal anomalies are not syntactically ambiguous (see above) but syntactically correct sentences, this possibility can be excluded. Hence, we are left with two scenario’s that could explain a mood by plausibility interaction for P600. First, mood could modulate language comprehension by increasing or decreasing the use of heuristics. As pointed out above, in the emotion literature it is generally agreed upon that happy mood leads to more reliance on heuristic processing than sad mood (see Bless & Schwarz, 1999 for a review). Following this, we would expect that happy participants are more inclined to rely on heuristics while reading implausible sentences (like the semantic reversals) than sad participants. The extent to which participants rely on a good-enough representation of the linguistic input will be inferred from the presence vs. absence of a P600 effect to the critical verb. The second scenario is that mood could modulate language comprehension by increasing or decreasing more general processes, like attention. More specifically, participants in a happy mood could pay more attention to the linguistic input, while people in a sad mood could pay less attention to the linguistic input. Early ERP components, in particular the P1(125–175 ms time-window) and the N1 (175–225 ms time-window) are taken to reflect perceptual and attentional processing, respectively (Mangun, 1995; Mangun & Hillyard, 1991). Possible effects of mood on these early perceptual and attention related components will be investigated since these effects can reveal an influence of attention. An effect of mood on attention should be accompanied by modulation of P1 and/or N1 as a function of mood. As will be discussed below, attention can also affect language relevant ERP components like the N400 and P600.

that reported no drug abuse, neurological, mental, or chronic bodily diseases, or medication for any of these. All were native speakers of Dutch, had no reading disabilities, were right-handed and had normal or corrected-to-normal vision. Hand dominance was assessed with an abridged Dutch version of the Edinburgh Inventory (Oldfield, 1971). Sixteen participants reported the presence of lefthandedness in their immediate family. One participant had to be excluded from the analyses because of excessive eye-movement artifacts. 2.2. Materials The semantic list consisted of 68 Dutch sentences with centrally embedded relative clauses. For each sentence a plausible and an implausible version were created, yielding a total set of 136 sentences (see Table 1). The semantically anomalous sentences expressed scenarios conflicting with general world knowledge (e.g., ‘foxes are not very likely to hunt poachers’ whereas ‘poachers are likely to hunt foxes’). The anomalies resulted from reversing the first and the second noun phrase of semantically acceptable sentences. The two noun phrases could both serve as the agent and the patient of the action expressed by the verb ending the relative clause (e.g., ‘foxes and poachers can hunt’ as well as ‘be hunted’). The anomaly was not evident before the relative clause’s verb. This was done to ensure that the detection of the anomaly required deep processing of the relative clause, in that it depended on the successful integration of the verb with both noun phrases. In half of the sentences, the two noun phrases had the same grammatical number and in the other half they had a different number (singular or plural). The two versions of each sentence were counterbalanced across lists. Each list contained 34 semantically plausible sentences and 34 semantically implausible sentences. In earlier studies, we showed that this type of sentences reliably elicits a P600 effect (Kolk et al., 2003; Vissers et al., 2007). Sixty-eight filler sentences were added to each list: 17 plausible right-branching sentences, 17 semantically implausible right-branching sentences (e.g., De rechter luisterde naar de beklaagde die opkwam voor zijn advocaat. Literal translation: The judge listened to the defendant who stood up for his lawyer.), 17 semantically plausible conjunctions and 17 conjunctions with a semantic implausibility (e.g., De zeehonden doken in het water en vingen de ijsbeer. Literal translation: The seals plunged into the water and caught the polar bear.). 2.3. Procedure

2.1. Participants

Participants were seated in a closed chamber. Sentences were presented in serial visual presentation mode at the center of a PC monitor. Word duration was 345 ms and the stimulus-onset asynchrony (SOA) was 645 ms. Sentence final words were followed by a full stop. The inter-trial interval was 2 s. Words were presented in black capitals on a white background in a 9 cm by 2 cm window at a viewing distance of approximately 1 m. Each sentence was preceded by a fixation cross (duration 510 ms) followed by a 500 ms blank screen. Participants were instructed to attentively read each sentence. No additional task was given. To make sure that the participants would actively read the sentences, they were told that questions would be asked about the sentences at the end of the experiment. The list was split up into three blocks; there was a pause between blocks, during which a film fragment was presented (see mood induction procedure) and each block was preceded by two filler items. Because eye movements distort the EEG recording, participants were trained to make eye movements, only in the period that the prompt was present (stimulus duration was 2295 ms). Prompt offset was followed after 705 ms by a fixation cross indicating the start of the next trial.

There were 30 female participants, all were students of the Radboud University (mean age¼ 22 years; age range ¼18–31). Recent evidence makes clear that the widespread assumption that subject gender matters little if at all in studies on the neurobiology of emotional memory is no longer tenable and, therefore, should be abandoned (Cahill, 2006). In line with this, preliminary evidence suggests that female and male participants process meaning differently in a positive vs. neutral mood (Federmeier, Kirson, Moreno, & Kutas, 2001). We therefore tested female participants in the present study. It remains to be seen whether the presented results generalize to a male population. Other criteria were as follows: only those participants were selected

2.3.1. Mood induction procedure Immediately before the EEG recording, the mood induction procedure was started. Between reading the experimental sentences for comprehension, mood was manipulated by presenting short film clips that either displayed fragments from a happy movie or a sad movie (see also Westermann et al., 1996). The happy movie fragments were cut from the Warner Brothers’ movie Happy Feet; the sad movie fragments were cut from the Universal Pictures’ second World War movie Sophie’s Choice. The clips used, showed unambiguous, unipolar emotions and affective situations, and participants were urged to use these clips to help them enter the specified mood. Participants were randomly assigned to the happy mood

2. Method

Table 1 Examples of the semantically plausible and implausible versions of the experimental sentences.

Experimental sentence Word-by-word translation Paraphrase

Plausible sentence

Implausible sentence

De docent die aan de studenten lesgaf kwam het lokaal in The teacher who on the students gave [singular] lesson entered the room The teacher who taught [singular] the students entered the room

De studenten die aan de docent lesgaven kwamen het lokaal in The students who on the teacher gave [plural] lesson entered the room The students who taught [plural] the teacher entered the room

C.Th.W.M. Vissers et al. / Neuropsychologia 51 (2013) 1027–1039 condition or the sad mood condition. Previous experiments indicated that these two movies are effective in inducing positive and negative mood, respectively (Chwilla et al., 2011; Vissers et al., 2010). The length of the film clips varied between 2.48 and 7.27 min, with a mean length of 5.41 min for the happy mood condition, and 5.47 min for the sad mood condition. In total we presented three film clips to facilitate the prolongation of the induced happy or sad mood throughout the entire experiment. Before and after watching each of the three film fragments, participants were asked to rate their own mood on a mood rating scale ranging from ‘extremely sad’ (  10) to ‘neutral’ (0) to ‘extremely happy’ (þ 10). Participants indicated their mood rating by moving a cursor on a keyboard to the level that corresponded with their current emotional state. There were two reasons why we manipulated mood between subjects. First, based on a pilot study, we were skeptical about switching on and off a positive vs. negative mood within a single recording session. Second, although it would have been possible to invite the same group of subjects to a second recording session this would have resulted in repetition of the critical stimulus materials. As it can take a long time before stimulus repetition to vanish (Cave, 1997) and given that late positivities are sensitive to stimulus repetition (Olichney et al., 2006), we preferred not to present the language materials twice.

1031

3. Results 3.1. Effectiveness of the mood induction procedure As Fig. 1 shows and supported by the statistical analyses reported below, the mood induction procedure effectively induced the intended mood. That is, after watching happy film clips, participants were significantly happier than at baseline measurement (all pso.05). Similarly, after watching sad film clips, participants were significantly sadder, than at baseline measurement (all pso.001). Furthermore, a significant difference in mood scores between the happy and sad mood condition was present after each of the film clips (all ps o.001). See Table 2 for the mean mood scores, presented separately for the participants who watched happy film clips and those who watched sad film clips. As evident from Fig. 1, there were no differences in mood scores before film clips were presented (F o2.5).

2.4. EEG data acquisition and analyses

3.2. Event-related potentials

EEG was recorded with 27 tin electrodes mounted in an elastic electrode cap (Electrocap International). The electrode positions included standard International 10–20 system locations over the left and right hemispheres at the frontal (F3, F4, F7 and F8), midline (Fz, Cz, Pz, Oz), parietal (P3, P4) and temporal (T5, T6) sites. Eight extra electrodes were placed at the frontal (F3A, FzA, F4A, F7A, F8A), midline (Oz) and parietal (P3P, P4P) sites. In addition, eight electrodes were placed at nonstandard electrode positions previously found to be sensitive to language manipulations (e.g., Holcomb & Neville, 1990): left and right anteriortemporal sites (LAT and RAT: 50% of the distance between T3/4 and F7/8), left and right temporal sites (LT and RT: 33% of the interaural distance lateral to Cz), left and right temporoparietal (LTP and RTP: Wernicke’s area and its right hemisphere homolog: 30% of the interaural distance lateral to a point 13% of the nasion–inion distance posterior to Cz), and left and right occipital sites (OL and OR: 50% of the distance between T5/6 and O1/2). The left mastoid served as reference. Electrode impedance was less than 3 Kohms. The electro-oculogram (EOG) was recorded bipolarly; vertical EOG was recorded by placing an electrode above and below the right eye and the horizontal EOG was recorded via a right to left canthal montage. The signals were amplified (time constant ¼8 s, bandpass ¼0.02–30 Hz), and digitized online at 200 Hz. Presentation of stimuli and recording of performance data was accomplished by a Macintosh computer. EEG and EOG recordings were examined for artifacts and for excessive EOG amplitude (4100 mV) extending from 100 ms before the onset of the critical verb ending the relative clause to 1000 ms following its onset. Averages were aligned to a 100-ms baseline period preceding the critical verb.

The waveforms are presented separately for each mood because the analyses revealed an interaction between plausibility and mood. The grand mean ERPs for the happy mood condition to the critical verbs are presented in Fig. 2, and those for the sad mood condition are presented in Fig. 3. As can be seen in these figures, the critical verbs elicited an early ERP response which is characteristic for visual stimuli—that is, an N1 followed by a P2, which at occipital sites was preceded by a P1. These components were followed by a broad negative wave in the 300–500 ms epoch, peaking at about 400 ms, the N400, which was largest at central and posterior sites. It is well known that each open class word elicits an N400 component (e.g., Kutas & Van Petten, 1994). Inspection of the waveforms for the happy mood condition suggests that no N400 effect (i.e., more negative-going amplitudes to semantically implausible than plausible scenarios) was present. For the sad mood condition, at some

2.4.1. Quantification of ERP components Based on previous studies using the same or similar materials (Kolk et al., 2003; Van Herten, Kolk, & Chwilla, 2005; Vissers et al., 2006), mean amplitudes were calculated in the 600–800 ms time window to capture P600 effects. We also checked for main effects of mood and possible mood by plausibility interactions for earlier ERP components preceding P600, in particular for the P1, N1 and the N400. The P1 was measured in the standard 125–175 ms time-window, and the N1 in the standard 175–225 ms time-window. Analyses were performed for the P1 and N1 because these components are generally taken to reflect perceptual and attentional processing, respectively (Mangun, 1995; Mangun & Hillyard, 1991). Supplementary analyses were performed for the N400 (measured in the 300– 500 ms time-window) which is highly sensitive to semantic processes (e.g., Chwilla, Kolk, & Mulder, 2000; Kutas & Federmeier, 2000), to investigate whether the mood manipulation affected semantic processing Repeated Measures ANOVAs were carried out on mean amplitude with plausibility (semantically plausible vs. semantically implausible), hemisphere (left, right) and site (11 levels with as left hemisphere sites F7a/F3a/F7/F3/LAT/ LT/LTP/P3/P3p/T5/OL; and right hemisphere sites: F8a/F4a/F8/F4/RAT/RT/RTP/P4/ P4p/T6/OR) as within-subject factors and mood (positive vs. negative) as a between-subject factor. The scalp distribution was further explored in two separate ANOVAs, one for the midline sites and one for the anterior vs. posterior regions of interest (ROI). The midline analyses included the additional factor site (Fza, Fz, Cz, Pz, Oz). For the ROI analyses a region of interest (anterior vs. posterior) by hemisphere by site (F7a/F3a/F7/F3 vs. P3/T5/P3p/OL vs. F8a/F4a/F8/F4 vs. P4/T6/P4p/OR) design was used. The estimated Greenhouse and Geisser coefficient epsilon was used to correct for violations of the assumption of sphericity. All reported p-values are based on corrected degrees of freedom.

Fig. 1. Mean mood rating scores ranging from  6 (indicating ‘‘sad’’) to þ 8 (indicating ‘‘happy’’) for the three film fragments comprising the mood induction procedure (MIP) separately for the participants assigned to the two mood conditions (happy mood condition vs. sad mood condition).

Table 2 Mean mood scores (SD) for the baseline measurement and after the first film clip, for both the happy mood condition and the sad mood condition.

Happy mood Sad mood

Baseline mood score

Mood score after first film clip

5.00 (2.7) 5.67 (1.8)

6.53 (2.1)  .73 (2.68)

1032

C.Th.W.M. Vissers et al. / Neuropsychologia 51 (2013) 1027–1039

F3A

FZA

F4A

F3

Fz

F4

LAT

Cz

RAT

P3

Pz

P4

P3P

Oz

P4P

-5

Voltage

0 [µV]

0

OL

200

400

600

800

Time

1000 [ms]

OR

Plausible

5 0

200

400

600

800

Time

Implausible

1000

0

[ms]

200

400

600

800

Time

1000 [ms]

Fig. 2. Grand ERP averages for the happy mood condition, time-locked to the onset of the critical verb superimposed for the two levels of condition for all midline sites and a representative subset of lateral sites. Negativity is plotted upwards.

Voltage

[µV]

-5

F3A

FZA

F4A

F3

Fz

F4

LAT

Cz

RAT

P3

Pz

P4

P3P

Oz

P4P

0

OL

200

400

600

Time

0

800

1000 [ms]

OR

Plausible

5

0 Time

200

400

600

800

1000 [ms]

Implausible

0 Time

200

400

600

800

1000 [ms]

Fig. 3. Grand ERP averages for the sad mood condition, time-locked to the onset of the critical verb superimposed for the two levels of condition for all midline sites and a representative subset of lateral sites. Negativity is plotted upwards.

C.Th.W.M. Vissers et al. / Neuropsychologia 51 (2013) 1027–1039

anterior sites a small increase in negativity within the N400 window appeared to be present for semantically implausible compared to plausible verbs. In the happy mood condition, the N400 was followed by a slow positive shift starting at about 500 ms and extending up to 1000 ms. The amplitude of the positive shift in the happy mood condition, seemed to be modulated by plausibility, with larger (i.e., more positive-going) amplitudes for semantically implausible verbs than for plausible verbs. The positivity in terms of timing and scalp distribution resembles the P600 elicited by syntactic anomalies (e.g., Hagoort et al., 1993; Osterhout & Holcomb, 1992; Osterhout & Mobley, 1995) and, more recently, to semantic anomalies (e.g., Kim & Osterhout, 2005; Kolk et al., 2003; Van Herten et al., 2005). In contrast, for the sad mood condition, at centroparietal sites that typically show the largest P600 activity, no positive shift was observed. Furthermore, in the standard P600 window (600–800 ms) no difference in mean amplitude was observed between the semantically implausible compared to the plausible critical items. Note that at anterior sites, mean amplitudes seemed to be more negative-going to semantically implausible than to plausible items—that is, the effect was in the opposite direction than that observed for the happy mood condition. 3.3. Statistical analyses About 5% of the trials were excluded from the analyses because of artefacts of which 3% belonged to the happy mood condition and 2% to the sad mood condition. The mean number and standard deviation of the number of trials for the semantically plausible condition and implausible condition were 28.63 (4.25) and 28.93 (4.07), respectively. A t-test confirmed that there were no differences in the number of trials across conditions, t (1,29)¼.56. 3.4. P600 window (600–800 ms) No effects of plausibility were found neither for the midline nor for the lateral sites or for the ROI analyses, all Fs o1.1. This indicates that across mood conditions, no differences in mean amplitudes between the semantically plausible and implausible items were found. Most importantly, two-way plausibility  mood interactions were obtained (midline sites: F(1,28)¼ 6.61, p o.02; all lateral sites: F(1,28)¼6.96, po.02; ROI: F(1,28)¼ 7.23, p o.02). In addition, three-way plausibility  mood  site interactions were present for the midline sites, F(4,112) ¼2.77, p o.05, and for the lateral sites, F(10,280)¼3.12, p o.04. Likewise, the ROI ANOVA revealed a plausibility  mood  ROI  site interaction F(3,84)¼5.02, p o.01. On the basis of these interactions separate analyses were performed for the different mood conditions. 3.4.1. Happy mood condition For the happy mood condition, the ANOVA for the midline sites revealed an effect of plausibility, F(1,14)¼6.46, p o.025, indicating that mean amplitudes were more positive-going for the semantically implausible than for the plausible items (1.53 mV vs. 0.39 mV, respectively). No plausibility  site interaction was obtained, F o1.1, indicating that the P600 effect for the happy mood condition was broadly distributed across the midline. In line with this single site analyses revealed a significant P600 effect at Fza, Fz, Cz, Pz and Oz (ps o.04). The ANOVA for the lateral sites yielded a main effect of plausibility, F(1,14)¼5.54, p o.04 and a plausibility  site interaction, F(10,140) ¼3.30, p o.02. Follow up single-site analyses revealed a P600 effect at bilateral anterior sites (F3a, F3, F8a, and F4a: pso.05) and

1033

posterior sites (P4, P3p and OL: ps o.05). The ROI ANOVAs yielded a main effect of plausibility, F(1,14)¼6.56, p o.03 and a plausibility  ROI  site interaction, F(3,42)¼3.50, p o.04, confirming the presence of a P600 effect for a subset of anterior and posterior sites. 3.4.2. Sad mood condition For the sad mood condition, the ANOVA for the midline sites did not yield an effect of plausibility, F(1,14)¼ 1.32; p ¼.27. Mean amplitudes for the semantically plausible verbs were a bit more positive than those for the implausible verbs (0.69 mV and 0.14 mV, respectively). The plausibility  site interaction was not significant, F(4,56)¼2.48; p ¼.093. The ANOVA for the lateral sites did not disclose an effect of plausibility, F(1,14)¼1.75; p¼ .21, a plausibility  site interaction, Fo1.75, or other interactions with plausibility that could have pointed at differences in mean amplitude between plausible and implausible verbs. Also for the ROI ANOVAs, no effect of plausibility, Fo1.5, plausibility  site interaction, F o1.1, or other relevant interactions were obtained. In sum, in contrast with the happy mood condition in which a P600 effect was present for implausible verbs, for the sad mood condition there were no indications for a modulation in mean amplitude in the P600 window. Visual inspection of the waveforms for the sad mood condition suggested the presence of an anterior negative effect to semantically implausible compared to plausible verbs. The plausibility  ROI interaction for the 600–800 ms epoch was not significant F(1,14)¼2.78, p ¼.118. Supplementary analyses for smaller timewindows (i.e., 600–700 ms and 700–800 ms) were performed to test the reliability of the anterior negative effect. The analyses for the sad mood condition for the 600–700 ms window did not disclose effects of plausibility, all p-values o.17. No other relevant interactions with plausibility were present. The analyses for the 700–800 ms window for the midline and the lateral sites did not yield effects of plausibility, all Fs o1.1, or plausibility  site interactions, Fso2.1. Also for the ROI analysis no significant plausibility  ROI interaction was present, F(1,14)¼ 3.51, p o.085. In brief, the analyses for the sad mood condition did not support the presence of a significant anterior negative effect to implausible vs. plausible verbs. 3.5. Results for early ERP components To check for possible mood by plausibility interactions for earlier components, analyses were also performed for the P1 and N1. A main effect of mood or plausibility by mood interaction for these early components would support scenario three according to which the effects of mood are caused by more general nonlinguistic factors. In particular, the results for these early ERP components inform us about a possible contribution of differences in early perceptual and/or attentional processes between participants in the happy vs. sad mood condition. The main results were as follows: for the P1 (125–175 ms) and N1 (175– 225 ms) no main effects of mood were present [midline sites (Fso1); lateral sites: (Fs o1)]. Likewise, no effects of plausibility were found [midline sites (Fso1); lateral sites (Fso2)]. Furthermore, no two-way interactions with mood and/or plausibility or other interactions with site and/or hemisphere that could have pointed at differences between conditions as a function of mood were present for the midline (Fso1.5) nor for the lateral sites (Fs o0.5). The ROI analyses for the N1 window yielded a trend towards a plausibility  ROI  hemisphere  site interaction, F(3,84)¼3.04, p¼.053. Follow up single sites analyses disclosed a reliable difference in mean amplitude with larger amplitudes for implausible than plausible verbs at one single site, RT: po.05.

1034

C.Th.W.M. Vissers et al. / Neuropsychologia 51 (2013) 1027–1039

No other effects were present.1 Based on these results we conclude that the effects of mood on P600 cannot be explained in terms of early perceptual and/or attentional processes. It has to be pointed out that later language-relevant components, in particular N400 and P600, have also been shown to be affected by attentional factors. Therefore, as we will argue in Section 4 we cannot rule out a possible contribution of more general factors like attention to the P600 plausibility by mood interaction. 3.6. N400 (300–500 ms) window For N400, no main effects of plausibility (all Fs o1) or plausibility  mood interactions were obtained (midline, lateral and ROI analyses: all Fso1.1). In addition, the midline analysis did not yield a plausibility by site interaction, Fo1. For the lateral sites a trend toward a plausibility  hemisphere  site interaction, F(10, 280)¼2.49, p ¼.059, was found. Follow up t-tests for the single sites, however, disclosed that for none of the single sites a significant plausibility effect was present, all p-values 4.29. The ROI analyses did not yield interactions that could have pointed at the presence of a reliable condition effect. 3.7. Correlation analyses To test whether the size of the P600 effect in the happy mood condition was indeed accompanied by changes in emotional state, supplementary correlation analyses were performed, with size of the P600 effect and mean mood rating (computed over three mood ratings per subject) as factors. The size of the P600 effect was computed by the difference in amplitude to plausible verbs and to implausible verbs; this difference score was computed for every single electrode and for every subject. The Bonferroni correction was used to adjust the p-values for multiple comparisons (27 electrodes) (Sankoh, Huque, & Dubey, 1997). The new alpha level was: p ¼.008. Correlation analyses for the subset of electrodes that showed a significant P600 effect revealed significant correlations between the mood ratings and the size of the P600 effect. More specifically as Fig. 4 shows, significant correlations were obtained for: Fza, Fz, Pz, P3p and P4 (rs 4.631, all pso.008). With correlations ranging from .631 up to .798 at least 40% up to 63% of the variation in size of P600 is accompanied by variations in emotional state. These analyses show that the happier the mood, the larger the P600 effect. See Fig. 5 for the modulation of P600 by mood at one right parietal site (P4). 3.8. Mood effects on semantic reversal anomalies vs. syntactic anomalies In two previous articles we directly compared the scalp distributions of the P600 effect to semantic reversal anomalies with that to subject verb agreement errors for the same groups of participants (for details see Kolk et al., 2003; Van Herten et al., 2005). Important for the present purposes, these topographical analyses did not reveal differences in the scalp distributions between the syntactic P600 effect and the semantic P600 effect [as indicated by the absence of a condition (syntactic vs. semantic anomaly) by plausibility interaction or relevant interactions with site and/or hemisphere that could have pointed at differences in scalp distribution between the two kinds of effects]. These analyses indicated that the two kinds of P600 effects were largest 1 Furthermore, no significant effects of plausibility and/or mood were obtained in the 0–100 ms, 100–200 ms, 200–300 ms and the 300–500 ms windows for midline and lateral sites (all Fs o 2) which cover the Early Left Anterior Negativity [ELAN] and Left Anterior Negativity [LAN] windows.

at central and posterior midline and lateral sites. The topographical analyses thus indicated that the scalp distributions of the syntactic P600 effect and the semantic P600 effect were similar. In the Vissers et al. (2010) study and the present article, the effects of a participant’s background emotional state on the processing of subject–verb agreement errors and semantic reversal anomalies were investigated, respectively. The main result of these two experiments was that mood modulated the P600 effect to syntactic anomalies and semantic reversal anomalies in a similar way. In the Vissers et al. study, this interaction observed for syntactic anomalies indicated a large P600 effect for the happy mood condition and a strong reduction in P600 effect for the sad mood condition (a P600 effect was only present at two posterior sites, that is at P3p and P4p). In the present experiment, the interaction observed for semantic reversal anomalies reflected the presence of a clear P600 effect for the happy mood condition vs. absence of a P600, and consequently of a P600 effect, for the sad mood condition. In sum, a similar but non-identical pattern of P600 effects as a function of emotional state was found for syntactic anomalies vs. semantic reversal anomalies. To check for reliable differences of the effect of mood on the syntactic P600 effect and the semantic P600 effect, supplementary global analyses for the P600 (measured in the 600 to 800 ms time window), with mood and anomaly type (semantic reversal vs. syntactic anomaly) as between-participant factors and condition (plausible vs. implausible in the present study and correct vs. incorrect in the Vissers et al. study) as within-participant factor were carried out. The crucial question was whether a mood  anomaly type  condition interaction would be obtained. A main effect of anomaly type, all Fs 412, pso.01, indicated that overall mean amplitudes were more positive to syntactic anomalies than semantic reversal anomalies. The global analyses revealed condition effects for the midline, F(1,57)¼7.36, p o.01, lateral, F(1,57)¼6.93, p o.02, and ROI analyses, F(1,57)¼6.82, po.02. The main effects reflected that mean amplitudes across anomaly type were more positive for implausible/incorrect items than for plausible/correct items. More importantly, mood  condition interactions were obtained (midline: F(1,57)¼11.53, po.002, lateral: F(1,57)¼13.63, p o.001, and ROI: F(1,57)¼ 14.76, p o.001). These interactions confirmed that different P600 patterns were found for the happy mood condition vs. sad mood condition across anomaly type (see below). Consistent with previous results (Kolk et al., 2003; Van Herten et al., 2005), no interaction between condition and anomaly type was found, all ps 4.11. Furthermore, no interaction(s) of the factors condition and anomaly type with site, ROI, and/or hemisphere was found, that could have pointed at reliable differences in P600 effects between the syntactic P600 effect and the semantic P600 effect, all p-values 4.09. Importantly, there was no indication for a three-way mood  anomaly type  condition interaction, all Fs o1. In addition, there were no other relevant four-way or five-way interactions that could have unveiled possible differences in the effects of mood on the two anomaly types. These global analyses support two main conclusions: first, that mood affected the processing of semantic and syntactic P600 effects in a similar way. Second, in line with previous studies no evidence was found for topographical differences in the P600 effects to syntactic and semantic anomalies. Based on the two-way condition  mood interactions separate ANOVAs were carried out for the two mood conditions. For the happy mood condition, all analyses yielded clear condition effects (midline: F(1,28)¼16.21, po.001; lateral: F(1,28)¼22.12, po.001; ROI: F(1,28)¼23.00, p o.001), in the absence of a condition  anomaly interaction, all p-values4 .11. No other interaction(s) that could have pointed at differences in P600 pattern for syntactic vs. semantic reversal anomalies was found,

C.Th.W.M. Vissers et al. / Neuropsychologia 51 (2013) 1027–1039

F7A F3a

-.655

1035

F8a F4A F8

F7 F3

-.631

F4

LAT

RAT

RT

Cz

LT

LTP P3 T5

RTP -.720

-.798 T6 P4P

-.697 OL

OR Oz

Fig. 4. Pearson correlations on an idealized head, looking down, nose at the top, between the size of the P600 effect and the mood ratings, for the happy mood condition for the set of electrodes that showed significant P600 effects.

Fig. 5. Mean mood rating is presented on the x-axis. Size of the P600 effect is presented on the y-axis. The scatter diagram shows strong relation between the P600 effect and emotional state at one right parietal site (r¼ .77, p o .05). The bestfitting regression line is also plotted.

condition effects in the opposite direction –that is, mean amplitude was more negative going for implausible/incorrect than plausible/correct F(3,87)¼12.74, p o.01. Separate ANOVAs for the anterior ROI revealed an effect of condition, F(1,29)¼5.08, po.04, supporting that for anterior sites mean amplitude was more negative-going for implausible/incorrect than plausible/ correct verbs. Furthermore, a condition  anomaly type  site interaction was present, F(3,87)¼3.40, p o.04. Separate analyses for syntactic anomalies disclosed a more negative amplitude for incorrect than correct verbs at one site, F8a: po.05. The analyses for semantic reversal anomalies revealed an increase in negativity for implausible than plausible verbs at three anterior sites of the left hemisphere, namely F7, F7a, and F3a: p-values o.05. Separate ANOVAs for the posterior ROI did not yield a condition effect or a condition  anomaly type interaction, both Fso1.65. No other interactions of the latter two factors with site and/or hemisphere were obtained, all Fs o1. In sum, the global analyses revealed a clear effect of condition across anomaly type for the happy mood condition. For the sad mood condition, the global analyses supported (i) a strong reduction in P600 effect for the sad mood condition, and (ii) the presence of a small anterior distributed negativity for implausible/incorrect verbs compared to plausible/correct verbs across studies.

4. Discussion all Fs o1.4. In contrast, for the sad mood condition there was no indication for a condition effect, all Fs o 1. Likewise, there was no indication for a condition  anomaly interaction, all Fso1.2. A condition  site interaction was found for the midline sites and for the lateral sites, F(4,116)¼5.13, po.01 and F(10,290)¼6.57, p o.01, respectively. These interactions reflected the presence of

Little yet is known about the relation between emotion and processes of language comprehension. Recently, heuristics have been proposed to play a major role in language comprehension next to syntactic algorithms (e.g., Christianson et al., 2010; Ferreira & Patson, 2007; Vissers et al., 2007). Heuristic processing

1036

C.Th.W.M. Vissers et al. / Neuropsychologia 51 (2013) 1027–1039

entails that language users do not always take all relevant information into account, syntactic as well as semantic information, but base their interpretation on a shallow—that is, goodenough representation of the linguistic input (e.g., Ferreira, 2003). The main goal of this article was to investigate whether a participant’s background emotional state affects processes of language comprehension, in particular the use of heuristics. While it has been demonstrated that mood affects syntactic processing (Vissers et al., 2010), to our knowledge the relationship between mood and heuristic processing in the domain of language has not yet been investigated. In the present article the interplay between mood and language comprehension was studied by testing for an interaction between mood and the P600 which has been shown to be sensitive both to syntactic well formedness and to semantic plausibility (see for reviews Kuperberg, 2007; Van de Meerendonk et al., 2009). Different emotional states (happy vs. sad emotional state) were induced by presenting short film clips from a happy movie or a sad movie while participants read sentences with semantic reversal anomalies (e.g., ‘the cat that fled from the mice’) or their plausible counterparts (e.g. ‘the mice that fled from the cat’). A necessary condition for investigating the relationship between mood and the use of heuristics in language comprehension is that the mood induction procedure was successful. The behavioral results indicate that this was the case. Participants were in a significantly happier mood after watching the happy film clips and they were in a significantly sadder mood after watching the sad film clips. Now we can address the question whether differences in emotional state affect heuristic processing in language comprehension. The main P600 results were as follows. For the standard window for P600 (600–800 ms) a significant interaction between semantic plausibility and mood was obtained for both midline and lateral sites. These two-way interactions reflected the presence of a broadly distributed P600 effect for the semantically implausible verbs in the happy mood condition vs. absence of a P600 and hence of a P600 effect to the same verbs in the sad mood condition. While in the happy mood condition, a P600 effect occurred across the midline and a subset of anterior and posterior electrodes, no hint for a P600 effect was present in the sad mood condition. Note that the global analyses for the sad mood condition (600–800 ms time window) revealed an effect in opposite direction to the P600 effect, namely, an increase in a small anterior distributed negativity for semantically implausible verbs vs. plausible verbs.2 This negative effect was restricted to the left hemisphere. The functional significance of this anterior negative effect is yet unclear and its robustness has to await future studies. Late anterior negativities have been related to working memory processes (e.g., Kluender and Kutas, 1993). Therefore, this negativity may reflect an increase in working memory demand for the semantically implausible relative to the plausible scenarios for the sad mood condition. Let us now scrutinize the plausibility by mood interaction in the happy mood condition. Correlation analyses were performed to verify whether the size of the P600 effect in the happy mood condition goes together with changes in emotional state. Importantly, the results of the correlation analyses demonstrate that the modulation in P600 amplitude reported in the present study was accompanied by significant changes in mood ratings. In particular, for the happy mood condition there was a strong correlation between the size of the semantic P600 effect and the mood scores, with an increase in the P600 effect with increasing happiness.

2 The fact that the plausibility effect and/or interactions with this factor did not reach significance when the results of the present study were analyzed alone is likely due to a lack of statistical power.

For the semantic P600 effect, significant correlations were present at three midline sites (Fza, Fz, Pz) and two lateral posterior sites (P3p, P4). Of interest here is that the correlations for the central posterior midline sites that typically show largest P600 effects, are significant (see Fig. 5). The correlations in the happy mood condition reveal that at least 40% up to 63% of the variation in size of the P600 effect was accompanied by changes in emotional state as reflected by the mood scores. The correlational analyses support the claim that a participant’s background emotional state is a major factor in explaining the presence vs. absence of the P600 effect after semantic reversal anomalies. The main conclusion to be drawn from the present experiment is that the processing of semantic reversal anomalies is modulated by emotional state. That is: while a broadly distributed P600 effect was present in the happy mood condition, sad mood led to the absence of P600 after semantic reversals. This interaction could be explained by two scenarios, one in terms of heuristics and one in terms of more general factors like attention. We will first elaborate on an interpretation of this interaction in terms of linguistic processes –that is, in terms of heuristic factors. After that we will discuss an alternative account, namely that the mood by plausibility interaction was caused by more general factors like attention. As pointed out earlier, while reading semantic reversal anomalies, a conflict between an expected representation based on heuristic processing and an unexpected representation based on algorithmic processing, triggers a P600 effect (Kolk et al., 2003; Vissers et al., 2007). Accordingly, because people in a happy mood exploit heuristics while reading semantic reversals, a conflict arises when an unexpected word is encountered. This conflict between the expected and the unexpected word triggers the brain to reanalyze the linguistic input to check upon processing errors, which is reflected in the present study by the widely distributed P600 effect for the happy mood condition. People in a sad mood on the other hand, process sentences more completely and with a greater focus on detail. Sad people are therefore not misled by their world knowledge/do not use heuristics while reading semantic reversals. It is therefore not necessary for participants in a sad mood to reanalyze the sentences after an unexpected word, which is reflected in the present study by the absence of a P600 effect in the sad mood condition. Put shortly, we propose that in the present study, the presence of a clear P600 effect in the happy mood condition reflects the use of heuristics, while the absence of the P600 effect in the sad mood condition reflects that sad participants do not use heuristics while reading semantically implausible sentences. The proposal that happy mood increases the use of heuristics when reading, while sad mood leads to a bottom-up, detailfocused style of language processing also confers with the socalled affect-as-information hypothesis (see Clore et al., 2001; Clore & Huntsinger, 2007; Schwarz, 2002). According to this hypothesis, mood and emotion influence judgments directly by serving as information regarding how one feels about the object of judgment. The basic idea is that affective cues of mood influence cognitive processes by serving as experiential and bodily information about the value of one’s thoughts and inclinations. Happy mood, hence, reinforces our default strategy, that is global, heuristic processing, while sad mood inhibits this tendency, leading to bottom-up, local processing with a focus on detail. Put differently, people infer from a positive state (I feel good) that the object of judgement must also be good and they therefore rely on heuristics. Given the methodological similarity between the Vissers et al. (2010) and the present study, a direct comparison of the effects of mood on the semantic P600 effect (present study) vs. the syntactic P600 effect (Vissers et al., 2010), is possible. In both experiments an interaction between mood and condition (plausibility vs. correctness in the present study vs. the Vissers et al.

C.Th.W.M. Vissers et al. / Neuropsychologia 51 (2013) 1027–1039

study) was found for P600. In the Vissers et al. study this interaction observed for syntactic anomalies indicated a large P600 effect for the happy mood condition and a strong reduction in P600 effect for the sad mood condition. A similar but nonidentical ERP pattern was observed in the present study, in which a semantic P600 effect was found only for the happy mood condition and not for the sad mood condition. If mood would have affected the processing of syntactic vs. semantic anomalies differently, this should have been reflected by a mood by anomaly type by condition interaction. The results of the global analyses on this point are clear: there was no indication for a critical threeway interaction (all Fso1). From this we conclude that mood influences the processing of syntactic anomalies and semantic anomalies in a similar way. As mentioned before, there were three different scenarios to account for the mood by condition interaction in the Vissers et al. (2010) study. That is, mood could have modulated the processing of syntactic anomalies (i) by increasing or decreasing syntactic processing, (ii) by enhancing or decreasing more general factors, like attention, or (iii) by increasing or decreasing the reliance on heuristics. Importantly, we propose that heuristics cannot only be based on semantic expectancy but also on syntactic expectancy. The P600 after linguistic violations is based on the degree of expectation set up by the context (Kolk et al., 2003; Van Herten et al., 2005, 2006; Vissers et al., 2007). It is a reader’s default expectation that sentences are grammatical (Coulson et al., 1998a, 1998b). After all, despite that language comprehension and speech are highly complex processes, grammatical comprehension errors and speech errors are very rare.3 Also when reading syntactically incorrect sentences, based on heuristic processing the expected syntactically correct interpretations of these sentences is accessed. Based on the expectation for a syntactically correct sentence a conflict will be triggered when readers encounter an ungrammatical inflection. We propose that a conflict between a syntactically correct, expected representation and a syntactically incorrect, unexpected representation of the verb triggers reanalysis to filter out possible processing errors. This reanalysis is reflected by the P600 (Van de Meerendonk, Indefrey, Chwilla, & Kolk, 2011). The modulation of the P600 to subject–verb agreement errors and semantic reversal anomalies by mood, therefore could both be due to the use of heuristics. Specifically, on the expectation of either a syntactically coherent content in case of syntactic violations, or on the expectation for a semantically (conceptually) coherent content in case of semantic reversal anomalies. Based on the emotion literature, we predicted that happy people would be more inclined to rely on heuristics than sad people. More specifically, people in a sad mood, process the sentences more systematically and with a greater focus on detail. It is therefore less necessary for participants in a sad mood to reanalyze the sentence after an incorrect verb inflection, which could be reflected in a strong reduction of the P600 effect. People in a happy mood on the other hand, are more inclined to use heuristics. Based on heuristics the reader expects a correct verb; a conflict arises when an unexpected incorrect/unexpected verb is encountered. This conflict between the expected and the unexpected verb inflection triggers the brain to reanalyze the linguistic input, which could be reflected by the widely distributed P600 effect observed for the happy mood condition. In the present study, the effect of mood on the processing of semantic reversal anomalies was investigated, to further determine the locus of the effects of mood on language. 3 For instance, Garnham, Shillcock, Brown, Mill, and Cutler (1981) found a total number of 191 slips of the tongue (including grammatical speech errors) in a text corpus of 200,000 words; this is no more than about one speech error per thousand words.

1037

Based on the fact that the sentences containing semantic reversal anomalies are not syntactically ambiguous, the modulation of P600 by mood cannot be explained in terms of syntactic processing. This leaves us with two scenarios that could underlie the P600 mood by plausibility interaction observed in the present article, one in terms of heuristic processing and one in terms of more general factors like attention. An interpretation of the mood by plausibility interaction framed in terms of heuristics has been spelled out above. We will now turn to an alternative account of the interaction in terms of more general non-linguistic factors. According to the second scenario, mood could have affected language processing via more general factors like attention. More specifically, people in a happy mood could pay more attention to read sentences and, hence, detect semantically implausible sentences, while people in a sad mood could pay less attention to read sentences and hence do not detect semantically implausible sentences. Attentional processes could affect the ERPs in two ways: attention could have either modulated early ERP effects (Mangun, 1995), or later ERP components, like N400 or P600. With respect to the early ERP effects: supplementary analyses for ERP components sensitive to perceptual and attentional processes (the P1 and N1, respectively) did not reveal main effects of mood or differential effects of mood on the processing of semantic reversal anomalies. Supplementary analyses for N400 which precedes P600 and is sensitive to semantic processing, showed that N400 was not modulated by mood. However, the interaction between mood and the processing of semantic reveral anomalies as tapped by P600 could still reflect attentional processing, as it has been show that the depth of processing, modulates both the amplitude of the N400 (Chwilla, Brown, & Hagoort, 1995) and of the P600 (Gunter & Friederici, 1999). Future studies are required to examine a possible contribution of more general factors like attention in bringing about the mood by language interaction for semantic reversal anomalies reported here as well as for syntactic anomalies (Vissers et al., 2010). In our lab, we are currently investigating the contribution of attention to the mood by language interaction by manipulating attention (comparing the ERPs in a shallow processing task with those in a deep processing task), in addition to the factor mood (comparing a happy mood condition with a sad mood condition: Verhees, Vissers, & Chwilla, in preparation). Another more general factor that could underlie the observed mood by plausibility interaction is motivation. In particular, it could be argued that differences in motivation give rise to the observed P600 pattern: people in a happy mood could be more motivated to read sentences which leads to reanalysis for the semantic reversal sentences, while people in a sad mood could be less motivated to read sentences and, hence, are not inclined to reanalyze implausible sentences. Interpreting the interaction between mood and processing semantic reversals in terms of attention or motivation would be consistent with a recent account of P600 effects to semantic reversal anomalies by Brouwer, Fitz and Hoeks (2012). In contrast with several P600 accounts that propose a (semantic) expectation/ heuristic-based route to explain the P600 effect to semantic reversal anomalies (Bornkessel & Schlesewski, 2008; Kim & Osterhout, 2005; Kuperberg, 2007; Van Herten et al., 2005, 2006; Vissers et al., 2007), Brouwer et al. (2012) suggest a single route model of language processing. The authors claim that this single route model can explain all semantic reversal anomalies without assuming an influence of heuristics. Therefore, we will describe this model in more detail. Brouwer et al. (2012) state that P600 reflects word-by-word construction or updating linguistic representations of what is being read. In line with this, P600 after semantic reversal anomalies can be taken to reflect increased effort in integrating the critical word with its prior context to form a coherent representation of meaning. Furthermore, the P600 is

1038

C.Th.W.M. Vissers et al. / Neuropsychologia 51 (2013) 1027–1039

assumed to be task-dependent and its amplitude increases when the reader tries hard to integrate the critical word, while it decreases when the reader tries less hard. Applying this model to the present data the broadly distributed P600 effect for the happy mood condition could reflect a strong effort to semantically integrate words, while the absence of a P600 for the sad mood condition could reflect the absence of such an integration effort. With the data at hand we cannot tease apart the effects of mood from the effects of attention and/or motivation. As discussed elsewhere (Chwilla et al., 2011; Vissers et al., 2010) the proposal that mood leads to quantitative differences or to a general attenuation of cognitive processes is not supported by the emotion literature. Conversely, there is firm evidence that differences in mood lead to qualitative different strategies and that mood dependent processing styles exist (e.g., Schwarz, 2002). It is clear from the above, that future studies are required to determine the relative contribution of linguistic processes (syntactic processes and heuristic processes) on the one hand and of more general factors (attention and motivation) to the observed mood by language interactions on the other hand. Importantly, the absence of an effect of mood and/or plausibility on N400 and earlier language-related components like ELAN and LAN is at odds with the idea of a general reduction/attenuation of cognitive processes (s., a decrease in attention and/or motivation) for the sad mood condition. Therefore in light of the emotion literature and a recent psycholinguistic proposal that heuristics play an important role in language (Ferreira, 2003), we propose sad vs. happy mood leads to qualitative differences in language processing as reflected in the present study by a bias towards using heuristics for happy participants.

5. Conclusion The reported interaction between a person’s emotional state and the processing of semantic reversal anomalies provides further support that mood affects processes of language comprehension. In addition, the results of the global analyses show that the effect of mood, as reflected by modulations in P600, on the processing of semantic reversal anomalies was similar to the effect of mood on the processing of syntactic anomalies. Further empirical and theoretical work is needed to gain a better understanding of the relation between mood and processes of language comprehension by shedding light on the mechanism(s) that mediate the effects of emotional factors on language.

Acknowledgments This research was supported by the Donders Centre of Cognition. The first and the final authors contributed equally to the research reported here. We are grateful to Herman Kolk for valuable comments on a previous draft of this article. We thank Marieke van Duuren for testing participants and Daniel Fitzgerald for enabling us to use the mood induction procedure in the presented experiment. We are grateful for the technical support of the ERG group, in particular Hubert Voogd, Pascal de Water and Jos Wittebrood. References Ashby, F. G., Turken, A. U., & Isen, A. M. (1999). A neuropsychological theory of positive affect and its influence on cognition. Psychological Review, 3, 529–550. Barton, S. B., & Sanford, A. J. (1993). A case study of anomaly detection: Shallow semantic processing and cohesion establishment. Memory and Cognition, 21, 477–487.

Bever, T. G. (1970). The cognitive basis for linguistic structures. In: J. R. Hayes (Ed.), Cognition and the development of language (pp. 279–362). New York: Wiley. ¨ Bless, H., Clore, G. L., Schwarz, N., Golisano, V., Rabe, C., & Wolk, M. (1996). Mood and the use of scripts: Does being in a happy mood really lead to mindlessness? Journal of Personality and Social Psychology, 71, 665–679. Bless, H., & Schwarz, N. (1999). Sufficient and necessary conditions in dual process models: The case of mood and information processing. In: S. Chaiken, & Y. Trope (Eds.), Dual process theories in social psychology. New York: Guilford. Bornkessel, I., & Schlesewski, M. (2008). An alternative perspective on ‘‘semantic P600’’ effects in language comprehension. Brain Research Reviews, 59, 55–73. Brouwer, H., Fitz, H., & Hoeks, J. (2012). Getting real about semantic illusions: Rethinking the functional role of the P600 in language comprehension. Brain Research, 1446, 127–143. Cahill, L. (2006). Why sex matters for neuroscience. Nature Reviews Neuroscience, 7, 477–484. Cave, C. B. (1997). Very long-lasting priming in picture naming. Psychological science, 8, 322–325. Chwilla, D. J., Brown, C. M., & Hagoort, P. (1995). The N400 as a function of the level of processing. Psychophysiology, 32, 274–285. Chwilla, D. J., Kolk, H. H. J., & Mulder, G. (2000). Mediated priming in the lexical decision task: Evidence from event-related potentials and reaction time. Journal of Memory and Language, 42, 314–341. Chwilla, D. J., Virgillito, D., & Vissers, C. Th. W. M. (2011). The relationship of language and emotion: N400 support for an embodied view of language comprehension. Journal of Cognitive Neuroscience, 23, 2400–2414. Christianson, K., Luke, S. G., & Ferreira, F. (2010). Effects of condition on structural priming. Journal of experimental psychology: Learning, Memory and Cognition, 36, 538–544. Clore, L. C., & Huntsinger, J. R. (2007). How emotions inform judgment and regulate thought. Trends in Cognitive Science, 11, 393–399. Clore, G. L., Wyer, R. S., Dienes, B., Gasper, K., Gohm, C., & Isbell, L. (2001). Affective feelings as feedback: Some cognitive consequences. In: L. L. Martin, & G. L. Clore (Eds.), Theories of mood and cognition: A user’s guidebook. Mahwah, NJ: Erlbaum. Converse, B. A., Lin, S., Keysar, B., & Epley, N. (2008). In the mood to get over yourself: Mood affects theory-of mind use. Emotion, 5, 725–730. Coulson, S., King, J., & Kutas, M. (1998a). ERPs and domain specificity: Beating a straw horse. Language and Cognitive Processes, 13, 653–672. Coulson, S., King, J., & Kutas, M. (1998b). Expect the unexpected: Event-related brain response to morphosyntactic violations. Language and Cognitive Processes, 13, 21–58. Erickson, T. A., & Matteson, M. E. (1981). From words to meaning: A semantic illusion. Journal of Verbal Learning and Verbal Behavior, 20, 540–551. Federmeier, K. D., Kirson, D. A., Moreno, E. M., & Kutas, M. (2001). Effects of transient, mild mood states on semantic memory organization and use: An event-related potential investigation in humans. Neuroscience Letters, 305, 149–152. Ferreira, F. (2003). The misinterpretation of noncanonical sentences. Cognitive Psychology, 47, 164–203. Ferreira, F., Bailey, K. G. D., & Ferraro, V. (2002). Good-enough representations in language comprehension. Current Directions in Psychological Science, 11, 11–15. Ferreira, F., & Patson, N. (2007). The good enough approach to language comprehension. Language and Linguistics Compass, 1, 71–83. Fodor, J. A. (1983). The modularity of mind. Cambridge, MA: Bradford Books. Garnham, A., Shillcock, R., Brown, G., Mill, A., & Cutler, A. (1981). Slips of the tongue in the London-Lund corpus of spontaneous conversation. Linguistics, 19, 805–817. Gasper, K., & Clore, G. L. (2002). Attending to the big picture: Mood and global versus local processing of visual information. Psychological Science, 13, 34–40. Gigerenzer, G., & Goldstein, D. G. (1996). Reasoning the fast and frugal way: Models of bounded rationality. Psychological Review, 103, 650–669. Gigerenzer, G., Todd, P. M., & ABC Research Group (1999). Simple heuristics that make us smart. New York: Oxford University Press. Gunter, T. C., Stowe, L. A., & Mulder, G. (1997). When syntax meets semantics. Psychophysiology, 34, 660–676. Gunter, T. C., & Friederici, A. D. (1999). Concerning the automaticity of syntactic processing. Psychophysiology, 36, 126–137. Hagoort, P., Brown, C. M., & Groothusen, J. (1993). The Syntactic Positive Shift (SPS) as an ERP measure of syntactic processing. Language and Cognitive Processes, 8, 439–483. Hoeks, J. C. J., Stowe, L. A., & Doedens, G. (2004). Seeing words in context: The interaction of lexical and sentence level information during reading. Cognitive Brain Research, 19(1), 59–73. Holcomb, P. J., & Neville, H. J. (1990). Semantic priming in visual and auditory lexical decision: A between modality comparison. Language and Cognitive Processes, 5, 281–312. Kim, A., & Osterhout, L. (2005). The independence of combinatory semantic processing: Evidence from event-related potentials. Journal of Memory and Language, 52, 205–225. Kimchi, R., & Palmer, S. E. (1982). Form and texture in hierarchically constructed patterns. Journal of Experimental Psychology Human Perception Performance, 8, 521–535. Kluender, R., & Kutas, M. (1993). Bridging the gap-evidence from ERPs on the processing of unbounded dependencies. Journal of Cognitive Neuroscience, 5, 196–214. Kolk, H. H. J., Chwilla, D. J., Van Herten, M., & Oor, P. J. W. (2003). Structure and limited capacity in verbal working memory: A study with event related potentials. Brain and Language, 85, 1–36.

C.Th.W.M. Vissers et al. / Neuropsychologia 51 (2013) 1027–1039

Kuperberg, G. R. (2007). Neural mechanisms of language comprehension: Challenges to syntax. Brain Research, 1146, 23–49 [Special issue]. Kuperberg, G. R., Holcom, P. J., Sitnikova, T., Greve, D., Dale, A. M., & Caplan, D. (2003). Distinct patterns of neural modulation during the processing of conceptual and syntactic anomalies. Journal of Cognitive Neuroscience, 15, 237–272. Kutas, M., & Federmeier, K. D. (2000). Electrophysiology reveals semantic memory use in language comprehension. Trends in Cognitive Sciences, 4, 463–470. Kutas, M., & Van Petten, C. K. F. (1994). Psycholinguistics electrified. Event-related brain potential investigations. In: M. A. Gernsbacher (Ed.), Handbook of psycholinguistics (pp. 83–143). San Diego, CA: Academic Press. Mangun, G. R., & Hillyard, A. A. (1991). Modulations of sensory-evoked brain potentials indicate changes in perceptual processing during visual-spatial priming. Journal of Experimental Psychology: Human Perception and Performance, 17, 1057–1077. Mangun, G. R. (1995). Neural mechanisms of visual selective attention. Psychophysiology, 32, 4–18. McDonald, M. C., Maryellen, C., Pearlmutter, N. J., & Seidenberg, M. S. (1994). Lexical nature of syntactic ambiguity resolution. Psychological Review, 101, 676–703. Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburgh inventory. Neuropsychologia, 9, 97–113. Olichney, J. M., Iragui, V. J., Salmon, D. P., Riggins, B. R., Morris, S. K., & Kutas, M. (2006). Absent event-related potential (ERP) word repetition effects in mild Alzheimer’s disease. Clinical neurophysiology, 117, 1319–1330. Osterhout, L., & Holcomb, P. J. (1992). Event-related brain potentials elicited by syntactic anomaly. Journal of Memory and Language, 31, 785–806. Osterhout, L., & Mobley, L. A. (1995). Event-related brain potentials elicited by failure to agree. Journal of Memory and Language, 34, 739–773. Sankoh, A. J., Huque, M. F., & Dubey, S. D. (1997). Some comments on frequently used multiple endpoint adjustments methods in clinical trials. Statistics in Medicine, 16, 2529–2542. St. John, M. F., & McClelland, J. L. (1990). Learning and applying contextual constraints in sentence comprehension. Artificial Intelligence, 46, 217–257.

1039

Schwarz, N. (2002). Situated cognition and the wisdom of feelings: Cognitive tuning. In: L. Feldman Barrett, & P. Salovey (Eds.), The wisdom in feelings. New York: Guilford Press. Trueswell, J. C., & Tanenhaus, M. K. (1994). Toward a lexicalist framework for constraint-based syntactic ambiguity resolution. In: C. Clifton, K. Rayner, & L. Frazier (Eds.), Perspectives on sentence processing. Hillsdale, NJ: Lawrence Erlbaum Associates. Verhees, M. W. F. T., Vissers, C. T. W. M., Chwilla, D. J. The relation between language, emotion and attention in language comprehension: Evidence from P600, in preparation. Van de Meerendonk, N., Kolk, H. H. J., Chwilla, D. J., & Vissers, W. M. (2009). Monitoring in language perception. Language and Linguistic Compass, 3, 1211–1224. Van de Meerendonk, N., Indefrey, P., Chwilla, D. J., & Kolk, H. H. J. (2011). Monitoring in language perception: Electrophysiological and hemodynamic responses to spelling violations. NeuroImage, 54, 2350–2363. Van Herten, M., Chwilla, D. J., & Kolk, H. J. (2006). When heuristics clash with parsing routines: ERP evidence for conflict monitoring in sentence perception. Journal of Cognitve Neuroscience, 18, 1181–1197. Van Herten, M., Kolk, H. H. J., & Chwilla, D. J. (2005). An ERP study of P600 effects elicited by semantic anomalies. Cognitive Brain Research, 22, 241–255. Vissers, C. Th. W. M., Chwilla, D. J., & Kolk, H. H. J. (2006). Monitoring in language perception: The effect of misspellings of words in highly constrained sentences. Brain Research, 1106, 150–163. Vissers, C. Th. W. M., Chwilla, D. J., & Kolk, H. H. J. (2007). The interplay of heuristics and parsing routines in sentence comprehension: Evidence from ERPs and reaction times. Biological Psychology, 75, 8–18. Vissers, C. Th. W. M., Virgillito, D., Fitzgerald, D., Tendolkar, I., Speckens, A. E. M., Oostrom van, I., et al. (2010). Mood affects syntactic processing: Evidence from P600. Neuropsychologia, 48, 3521–3531. Westermann, R., Spies, K., Stahl, G., & Hesse, F. W. (1996). Relative effectiveness and validity of mood induction procedures: A meta-analysis. European Journal of Social Psychology, 26, 557–580.