In The Cognitive-Emotional Brain, Pessoa (Reference Pessoa2013) effectively argues against the traditional view on which interactions between emotion and cognition in the brain are interpreted exclusively in antagonistic terms. According to this traditional view, emotional and cognitive brain systems are distinct; they compete for processing resources and lead to mutual suppression. Pessoa argues that this view is restrictive and does not capture the many possible ways in which emotion and cognition actually interact in producing behavior. His view, based on accumulating evidence that emotion and cognition share brain systems (especially those identified as informational hubs, e.g., amygdala, medial prefrontal cortex, anterior insula), is that emotion and cognition can interact in ways that result in more efficient behavior.

I agree with the gist of this argument. My aim here is to relate this account to complementary research on affective states and cognitive processes that are not considered in the book and that can strengthen Pessoa's position. One interesting case, for example, lies in the effect of mood on language comprehension, especially at the discourse and sentence level.

Pessoa's account is based mainly on research about processing emotional stimuli, focusing primarily on visual perception and executive functions and touching upon trait anxiety and depression. The cases Pessoa considers are therefore those in which emotion has a strong, rapid bottom-up effect (emotional stimuli) or has a persistent and prolonged top-down effect (trait anxiety and depression). Research on the effects of experimentally induced happy and sad moods on cognitive processes has shown that moods can influence performance when they are preexisting and unrelated to the task at hand and, most importantly, when people are unaware of experiencing them (for a review, see Martin & Clore Reference Martin and Clore2001). These features put mood in the position to exert an influence on cognitive processes for longer periods than the emotions elicited by experimental stimuli (such as a shock or a disgusting picture). Mood's effect also differs from that of emotional stimuli in that it is top-down. At the same time, it differs also from the top-down effect of trait anxiety and clinical depression: Because mood waxes and wanes easily in everyday life, it has less extreme and persistent effects on cognition. We discuss here research on the effects of mild happiness and mild sadness, as examples of moods that listeners and readers may be experiencing during language comprehension.

Experimentally induced moods can have a strong effect on a wide range of high-level cognitive functions (e.g., judgment, memory, executive functions; for reviews, see Clore & Huntsinger Reference Clore and Huntsinger2007; Martin & Clore Reference Martin and Clore2001). The fact that mood can influence integration processes during discourse comprehension is a recent finding; language has been usually studied as one of the cognitive functions most impermeable to affect, at least at the sentence and discourse level.

Recent research has found that incidental mood can have a strong and rapid influence on the way discourse is understood. Incidental mood is elicited with a procedure of which participants are not aware and that results in mild affective changes escaping participants' awareness (e.g., Koch et al. Reference Koch, Forgas and Matovic2013; Matovic et al. Reference Matovic, Koch and Forgas2014). Behavioral evidence has shown that readers induced into a happy or sad incidental mood judge a positive or a negative story ending as more surprising when its valence mismatches that of the mood (Egidi & Gerrig Reference Egidi and Gerrig2009). This is seen not only in explicit behavioral evaluations after reading, but also in neural responses during listening to the story endings (Egidi & Nusbaum Reference Egidi and Nusbaum2012). Mood interacts with linguistic content as early as 400 ms after the presentation of a critical word. The study by Egidi and Nusbaum (Reference Egidi and Nusbaum2012) examined the effect of happy, sad, and neutral moods on the comprehension of a positive and negative ending that concluded a very short story. Mood modulated the auditory N400 effect, with a pattern of increased N400 for story endings mismatching the valence of the mood (for consistent results; see also Chung et al. Reference Chung, Tucker, West, Potts, Liotti, Luu and Hartry1996). The N400 pattern for the happy and sad group further differed from that of the neutral mood group. Importantly, the effect of mood in evaluations and N400 effect occurred when positive and negative endings were equally consistent and plausible (Egidi & Gerrig Reference Egidi and Gerrig2009; Egidi & Nusbaum Reference Egidi and Nusbaum2012). A study by van Berkum and colleagues (Reference van Berkum, De Goede, Van Alphen, Mulder and Kerstholt2013) further found a reduced N400 for sad mood (vs. happy mood) after verbs that create an implicit causality bias (e.g., Carice annoyed Tom because he/she repeatedly had to correct her mistakes). At the sentence level, Federmeier et al. (Reference Federmeier, Kirson, Moreno and Kutas2001) also examined the effects of positive and neutral moods on semantic organization by using sentence pairs. In this study, positive mood, as compared to neutral mood, was associated with a decrease in N400 amplitude between unexpected items of different categories (e.g., They wanted the hotel to look more like a tropical resort. So along the driveway they planted rows of pines/tulips).

Taken together, these results highlight a very direct effect of mood on cognition, with no intentional component. In these cases, the effect of mood cannot therefore be attributed to motivational or attentional differences. Consistent with Pessoa's position, the influence of incidental mood suggests that the concepts of selection and competition for resources cannot easily explain the interaction between emotion and cognition, and that low arousal is not necessarily linked to decrement in performance. Instead, the impact of mood may be that of adjustment, as in the case of responding to positive and negative endings as more or less consistent depending on mood (Egidi & Gerrig Reference Egidi and Gerrig2009; Egidi & Nusbaum Reference Egidi and Nusbaum2012). However, general interference cannot be completely ruled out: Results such as the mood-dependent reduced sensitivity to implicit causality (van Berkum et al. Reference van Berkum, De Goede, Van Alphen, Mulder and Kerstholt2013) suggest that interference is a possible effect of mood on language.

At the sentence level, recent behavioral research has shown that incidental moods influence participants' ability to discriminate between ambiguous and unambiguous sentences (Matovic et al. Reference Matovic, Koch and Forgas2014). Other studies have also found evidence of mood modulations of linguistic processes that typically elicit N400 and P600. These studies have capitalized on a known tendency of happy and sad moods to promote different processing strategies. Several behavioral studies of social judgment and memory have in fact found that happy mood promotes a more global, top-down and heuristics-based type of processing, whereas sad mood promotes more local, bottom-up, and analytic type of processing (e.g., Bless Reference Bless and Forgas2000; Clore & Huntsinger Reference Clore and Huntsinger2007; Fiedler Reference Fiedler, Martin and Clore2001). Importantly, however, this effect is more often obtained when participants are induced to experience a certain mood by taking active part in changing the way they feel and maintaining it over time. Typically, participants are presented with a happy or sad stimulus (or are asked to think of something happy and sad) and are instructed to allow the stimulus to put them in a certain mood and to maintain that mood throughout the experiment (e.g., Chwilla et al. Reference Chwilla, Virgillito and Vissers2011; Vissers et al. Reference Vissers, Virgillito, Fitzgerald, Speckens, Tendolkar, van Oostrom and Chwilla2010; Reference Vissers, Chwilla, Egger and Chwilla2013). In this way, mood is voluntarily constructed and therefore likely to influence cognition through a different avenue than incidental mood. Mediating factors such as (1) awareness of being in a certain mood, (2) the strategies participants adopt to maintain a certain mood, (3) the potentially different ease with which different moods can be maintained, and (4) the constant attention participants pay to how they feel may result more easily in the adoption of different processing strategies than does incidental mood.

In the literature on sentence processing, this indirect effect of mood has been studied on the consistency effect elicited by reading high and low cloze-probability sentences (e.g., the pillows are stuffed with feathers/books; Chwilla et al. Reference Chwilla, Virgillito and Vissers2011). It was found that the N400 effect was differently distributed depending on participants' mood. Vissers and colleagues (Reference Vissers, Virgillito, Fitzgerald, Speckens, Tendolkar, van Oostrom and Chwilla2010; Reference Vissers, Chwilla, Egger and Chwilla2013) also found that the effect of constructed mood can modulate syntactic processing, as seen by a different amplitude and scalp distribution of the P600 in reading subject–verb disagreements and apparent syntactic anomalies (as a result of unusual semantic context: e.g., the fox the poachers hunts/hunt). Interestingly, no effects of mood on syntactic anomalies was found when a less explicit mood manipulation was used (van Berkum et al. Reference van Berkum, De Goede, Van Alphen, Mulder and Kerstholt2013), thus suggesting that some aspects of linguistic processing may be permeable to mood only when participants are made consciously aware of their mood via mediating factors (such as those mentioned above). In general, motivational and attentional explanations of the kind discussed by Pessoa are definitely applicable to the research on the effect of constructed mood on language.

With respect to the neural architecture underlying the effect of mood on discourse and sentence comprehension, there is only one study on this issue: an fMRI experiment examining the effects of incidental happy and sad mood on consistent and inconsistent story endings (Egidi & Caramazza Reference Egidi and Caramazza2014). The study's results are consistent with the notion that mood and language interact to create emergent activation patterns rather than attenuating or increasing activity in fixed networks. Specifically, a contrast between story endings that were consistent or inconsistent with prior context showed that happy and sad mood moderated inconsistency detection, but did so in different networks. For example, happy mood increased sensitivity to inconsistency in regions often linked to language comprehension, whereas sad mood increased sensitivity to inconsistency in other networks, less frequently linked to language processing. With respect to Pessoa's position, these results show that incidental mood can fundamentally alter the baseline functioning of linguistic processes, by restructuring the brain networks that perform that function. In this sense, these results are consistent with Pessoa's view of emotion-cognition relation as integration.

To conclude, research on the effects of mood on discourse and sentence comprehension shows that mood does not only interact with language processing in terms of interference or enhancement of function, but that it also fundamentally alters how linguistic processes are performed. This is especially seen in modulations of fast ERP responses to inconsistencies and linguistic anomalies and in mood-dependent reorganizations of the brain networks that perform linguistic functions. These results constitute a constructive addition to Pessoa's account, as they both increase its scope and highlight an additional way of emotion-cognition interaction that is not mentioned in the book.