Functionality of the N400 component and its application in studies of figurative language processing: a systematic review

Electrophysiological studies have increased in the field of Neuroscience of Language as they allow for detailed analyses of the temporal course of language processing. In figurative language research, the most employed technique is Event Related Potentials (ERPs). The N400 is the most studied ERP component and is usually described as a marker of difficulty or effort in semantic processing. However, many controversial results about N400 have been found. Two main reasons can be raised for these results: (1) methodological difficulties in controlling psycholinguistic variables that are known to affect language processing and (2) different interpretations regarding what the N400 indexes. In this systematic review our main goals are: (1) providing a general panorama of studies on the N400 component applied to metaphoric and idiomatic language and (2) discussing the different views adopted by researchers regarding functional N400 theories and how ERP studies of figurative processing fit into each of those theories. We found two main interpretations about the process which is indexed by the N400: N400 as a marker of lexical access or as a marker of semantic integration. We also found that this dichotomy has an impact on the choice of models to explain the results about figurative language processing.


Introduction
Figurative language is defined as the use of a word in a phrase or expression in an alternative meaning or sense in relation of literal use, such as "it's raining cats and dogs" or "He broke my heart". This is an important topic in language research due to its common occurrence in everyday language as well as the fact that it persists as a universal manifestation across different cultures (LAKOFF, 1979;JONHSON, 1980). These notorious aspects of figurative language lead to the development of important theories in Philosophy, Psycholinguistics, Cognitive Psychology and Neuroscience of Language on how humans process these types of stimuli. Grice (1975) and Searle (1979) proposed that we need additional cognitive mechanisms to process figurative language compared to literal meanings. At the same period, Lakoff and Jonhson (1980) proposed that figurative language is more than a simple linguistic tool. In their view, the nature of mind is metaphorical, which implicates the existence of a general mechanism to access meaning in a figurative way.
The impact of these theories led to the emergence of research focusing on figurative processing in the brain. Since 1980, several models have been proposed to explain how the brain processes figurative sentences using different research methods. The main goal has been to explore differences between com-

206
figurative processing. In the first view, figurative language has higher processing costs than literal processing and involves more neural resources or pathways. In the Direct View Hypothesis , it is proposed that, with sufficient context, figurative language is processed similarly as literal language (GIBBS, 2001). Finally, the Graded Salience model proposes that the pattern of processing of figurative content is also influenced by other factors that facilitate the comprehension (GIORA, 1999(GIORA, , 2002. Reaction time has been the main quantitative measurement applied and investigated in figurative language studies. Despite being very informative, this complex measure requires careful interpretation, given that it involves additional processes such as perceptual and motor components (FEDERMEIER, 2014). Particularly for figurative language, studies based on measuring reaction times may lead to incomplete conclusions since they are less sensitive to be mapped into a complex set of variables (FEDERMEIER, 2014).
To study brain responses more precisely and complement the reaction time data, there has been an increase of research on the past decades that combine electrophysiological measurements analyzed in combination with reaction times (THOMA; DAUM, 2006).
Despite the numerous methods in electrophysiology, event related potentials (ERP) has been the most employed procedure on language studies (SA-NEI; CHAMBERS, 2007). This technique exploits the optimal temporal resolution of EEG to provide an electrical pattern related to stimuli presentation (LUCK, 2014;SANEI;CHAMBERS, 2007). Generally, the ERP analysis considers the polarity and amplitude of electrical peaks following stimulus onset. Depending on the aim of the study, additional analyses of latency and topography can also be considered (LUCK, 2014).
Some peaks -the so-called ERP components -are repeatedly found for particular types of stimuli. The most well-known component in language literature, the N400, is described as a deflection with a peak close to 400 milliseconds classically related to anomalous association at the last word of a sentence (FEDERMEIER, 2014). The N400 was first described by Kutas e Hillyard (1980), starting a long series of articles reporting the N400 in linguistic stimuli related to dissociation in word pairs, anomalous sentence finalization, violation in high constraint contexts and unexpected words (FEDERMEIER, 2014). Beyond language, the N400 was also found in numerical cognition, face and object recognition paradigms tasks and semantic memory (FEDERMEIER, 2014).
The N400 is possibly the most well studied ERP in language and several studies have shown its relation to different aspects of language comprehension (for a review see FEDERMEIER, 2014;HILLAYARD, 1984;HAGOORT;LEVIN-SON, 2004;LAU;PHILLIPS;POEPPEL, 2008). Here, we will focus on how the N400 has been used to investigate figurative processing. Currently, there are two main views about what N400 indexes within this framework: (I) an integrative account, in which the N400 is depicted as representing a relatively late processing due to semantic integration of critical words with their respective previous context (FEDERMEIER, 2014;LAU et al., 2008); (II) a lexical account, in which the N400 is a marker of lexical retrieval from memory in addition to previous context (FEDER-MEIER, 2014;LAU et al., 2008).
In order to clarify these issues, the current review approaches the different information conveyed by the N400 in recent figurative processing studies. Our main objectives are: (1) to show a general panorama of studies on the N400 component applied to metaphoric and idiomatic language, summarizing types of task regarding modality (visual or auditory), language demands of the task, form of analyses (spatial or temporal), conditions, sentence extension, context support and control of independent variables (psycholinguistic factors); (2) to discuss the different views adopted by researchers in the field regarding functional N400 theories and how ERP studies of figurative processing fit to each theory.

Method
We conducted a systematic review of ERP studies in figurative language processing. The search was performed on two databases in January 2018: PubMed and the National Library of Medicine of the United States of America. The search was guided by the following keywords, using "AND" only as a boolean operator: "ERP AND metaphor"; "N400 AND metaphor"; "ERP AND idiom"; "N400 AND idiom"; "ERP AND figurative language"; "N400 AND figurative language".
No specific time-window was established for the database search. We excluded studies that did not use N400 data in debates on how the brain processes figurative stimuli, as well as studies using fMRI, PET, CT, NIRS experiments (except if concomitant with EEG) were. Apart from that, we excluded studies on irony, proverbs, metonymy and other types of figurative language that do not include metaphors and/or idiomatic expressions to restrict the research and discussion to relative similar figurative expression structures (idioms and metaphor expression can build differing only aspects as familiarity). Studies that used figurative language to analyze other aspects of language and cognition such as bilingualism, gesture processing, psychiatric syndromes, philosophy, and religion were also excluded. Finally, we excluded all works that did not show a clear ERP methodology or minimal aspects of the procedure.

Results
After organization and removal of all duplicates, 69 different articles were selected by title and abstract. We excluded 38 after reading the full text as they did not correspond to the criteria stated in methodology (Table 1) In almost all of these studies, visual stimuli were displayed on a monitor; only two studies were performed with auditory stimuli (LORUSSO et al., 2015;WEI-LAND;BAMBINI;SCHUMACHER, 2014), with one of them using masked prime word on monitor and sentences in earphones in a cross-modal manner (WEI-LAND et al., 2014).
In the majority experiments, stimuli consisted of sentences which were presented word-by-word with variable intervals between words specific to each experiment. This paradigm is named Rapid Serial Visual Presentation (RSVP), in which stimuli are presented serially and the volunteer makes the decision at the end of sequence. In some studies, however, stimuli were presented as entire sentences followed by a target sentence (LAURENT, 2006;RUTTER et al., 2012;SO-TILLO et al., 2005) The stimuli (sentences) held different extensions and combinations of conditions across studies. The most frequent comparisons were between conventional and unconventional metaphors and between literal and anomalous metaphors. But there were also experiments that used idiomatic stimuli in comparison with literal and anomalous sentences (see Table 3).
The studies reported different sets of measurements of psycholinguistic features of stimuli and conditions (Table 3). The most common variable registered (due to its possible effect on N400 amplitudes) was the participants' familiarity with the stimuli. This feature was accessed using psycholinguistic scales of familiarity, conventionality, frequency, and cloze probability (obtained from Likert scales in each study or linguistic databases). The cloze probability is an important factor associated with N400 and is often mentioned in N400 linguistic experiments. This concept refers to the probability of a sentence or string of words to be completed or followed by a specific word. For example, in the sentence "The dentist recommends brushing your teeth twice a ____" the probability of the "day" is much higher than "year" to fill the targeted gap (BLOCK.; BALDWIN, 2010).
Also, several experiments controlled variables such as: word-length of crit-

210
ical words (number of characters or syllables), sentence length (number of words in a sentence) and grammatical-and orthographical category of the stimuli (variables were summarized in Table 3).
In general, the studies analyzed the differences between N400 amplitudes across conditions (summarized in Table 4). The N400 time window considered for analysis across all studies shows some variability; different time windows were found across studies, with a majority of nine studies working with a time-window of 300-500 milliseconds (COULSON; VAN PETTEN, 2002;FORGÁCS et al., 2015;LU;ZHANG, 2012;SCHNEIDER et al., 2014;CURRAN, 2013;VESPIGNANI et al., 2010;ZHAO et al., 2011;ZHOU;ZHOU;CHEN, 2004). In addition, various studies used other ERP components as further evidence to make inferences regarding figurative comprehension. The Late Positive Component (LPC) and the P600 were the components most commonly used (summarized in Table 4).
Almost all studies recorded behavioral measurements in their tasks, only a few did not register behavioral responses (LU; ZHANG, 2012;MA et al., 2016;PYN-TE et al., 1996;SOTILLO et al., 2005;TARTTER et al., 2002;YANG;XUE, 2011).Most studies measured accuracy (ACC), reaction time (RT) or both (Table 4). Temporal analyses on the N400 were performed in all studies, generally investigating amplitudes to each type of stimulus.
This pattern also occurred when the metaphor condition was divided into conventional and unconventional metaphors, this difference is linked with how frequent a given metaphor is used. Some studies, have shown differences between variations or categories of metaphors, e.g. conventional metaphors having a smaller N400 negativity than novel ones (ARZOUAN et al., 2007a(ARZOUAN et al., , 2007b. Likewise, there are studies that did not find differences between novel, conventional and anomalous metaphors (LAI; CURRAN; MENN, 2009).
In relation to the N400 evoked by idiomatic expressions it was found that type of stimulus can evoke even less negativity than literal sentences (LAURENT et al., 2006;STRANDBURG et al., 1997). Only one experiment did not find this difference (PROVERBIO et al., 2009). Additionally, it was found the more lexicalized the idiom is, the less negativity it evokes.
Concerning these results about metaphors and idioms, it is important to remember that they are attached to specific models. The literature references used to support and explain their results were based on a variety of models. In general, indirect views of figurative processing were mainly considered in experiments. Basically, a direct view and indirect view of figurative language processing defends differences concerning to phases and time of processing. The direct view of figurative language makes an appoint that figurative language have no difference relative to literal one to be processed and indirect view argues that figurative processing has different phases do be processed in relation of literal sentence. In indirect view, that being said, mapping approaches were the most frequently mentioned models (ARZOUAN et al., 2007a(ARZOUAN et al., , 2007bCOULSON, 2007; COULSON; VAN PETTEN, 2002;LAI et al., 2009;LU;ZHANG, 2012;RUTTER et al., 2012;SCHNEIDER et al., 2016;CURRAN, 2013;WEILAND et al., 2014;ZHAO et al., 2011).
There are, however, citations of other models such as Graded Salience Hypothesis (DE GRAUWE et al., 2010;GOLDSTEIN et al., 2012;LAURENT et al., 2006) or Configuration Hypothesis (VESPIGNANI et al., 2010) for idioms and also -but less frequently -authors that support a direct view about figurative processing.
Lastly, it is important to mention that not all authors explicitly chose or mentioned a model; there are also studies that raise discussions with other aims, e.g.
describing the profile of different clinical subgroups without any specific concern on cognitive models about figurative language. These models will be better dis-

212
cussed in next sections of this review.

Discussion
We conducted a systematic review of the literature aimed at characterizing the N400 results for metaphor-and idiom processing in light of current functional theories of the N400. To achieve this, we summarized the experimental designs and methodology of those studies and examined their results involving N400 markers.
Our search included studies published up to January 2018, and as we have not determined a specific period for our review, it was possible to identify those studies using EEG to investigate metaphor and idiom processing have begun relatively recently (in the 1990's) and are still scarce (only 32 studies). In addition, most experiments were conducted in European or North American Universities in a reduced number of languages (8 idioms).

General characteristics of ERP figurative studies
The majority of experiments were performed in Germanic languages, less than one third were conducted in Roman languages. In almost one third of studies, the stimuli were presented in Hebrew and Chinese (see Table 2). Above all, it is worth mentioning that no studies were conducted in Portuguese, the second major Roman language. Nonetheless, in our review we found no specific N400 pattern that could be related to the language used in the studies.
Regarding stimuli presentation the majority of studies used visual tasks, such as word-pairs or sentences (presented word-by-word) and subjects were asked to make semantic judgments of the stimuli (i.e. semantic associations or sentence semantic plausibility). It is important to note that these two types of stimuli (word-pairs vs. sentences) differ not only in extension but also in the quantity of context offered for critical words, which is known to affect the comprehension process. For sentences, the cloze probability is currently considered the measure that best depicts the role of the context in predicting the critical word.
On the other hand, word pair studies require measures of semantic association between the two items. Whereas most sentence studies informed the cloze probability of the critical word, only a few word-pair studies measured semantic associations (see Table 3). Moreover, word pairs can have different uses according to the language (e.g. in Hebrew some word-pairs can be considered a single expression - Arzouan and Faust, 2000).
We found two studies that used auditory stimuli. Lorusso et. al. (2015) per-

218
formed a purely auditory task and Weiland and colleagues (2014) mixed visual and auditory presentation in their experiment. It is well known that the N400 is elicited by both modalities (visual and auditory) and that both forms elicit similar responses. Nevertheless, it is important to highlight that upon auditory presentations, the N400 can appear earlier and tends to be more frontal and less right biased (concerning the topography of activation) compared to the "classical" N400 found in visual tasks (FEDERMEIER, 2014;LAU et al., 2008).
Most studies compared three conditions in their experiments: "Literal x Metaphorical x Anomalous" (see Table 3). Some variations were found such as conventional vs. unconventional and familiar vs. unfamiliar metaphors. Those different designs allow analyses of specific linguistic aspects and their influence on N400 patterns. The use of conditions such as anomalous and literal sentences, conventional and unconventional metaphors identifies factors such as conventionality of a sentence and elucidates how this affects the N400 component and figurative processing in general (ARZOUAN et al., 2007a;GOLDSTEIN et al., 2012;LAI et al., 2009). The same holds for the cloze probability (LAURENT et al., 2006) and familiarity manipulations (DE GRAUWE et al., 2010). Also, cloze probability has been one of the main psycholinguist aspects researched in relation to N400, due to the strong negative correlation between the two measures (r=-0.9, for a review see FEDERMEIER, 2014), representing a measure from the degree of constraint of a word in a specific context (FEDERMEIER, 2014 (BAMBINI et al., 2016). The absence of this pattern gives other valuable insights that we will discuss in the following section (BAMBINI et al., 2016).

N400 as a marker in figurative models
Virtually all studies of this review have shown differences in the N400, with a stronger N400 for figurative processing than for literal sentences. This is commonly used as evidence to support indirect views on figurative processing, as these dissimilarities are understood as "proof" of different steps in the compre- This N400 difference between anomalous and metaphorical sentences is particularly important because it cannot be supported either by the Standard Pragmatic Model (whose assumptions have been weakly supported by electro- In our sample of studies, we found minimal variations across stimuli conditions. Most studies controlled sentences or target words on psycholinguistic patterns such as familiarity, cloze probability, conventionality, concreteness, and figurativeness. All of these factors are known to influence reaction times in comprehension tasks and also the N400 amplitudes, which results in different levels of this ERP component (DE GRAUWE et al., 2010;FEDERMEIER, 2014;LAU et al., 2008).
The most cited models in the investigated studies are mapping theories that encompass the graded pattern and these confounding factors (psycholinguistic variations of stimuli). These theories propose that a same system is used to understand figurative and literal concepts, but the comprehension process involves mapping the attributes from the target word to context information in order to establish similarities (COULSON; VAN PETTEN, 2002). In these explanations, N400 amplitudes are a proxy of the amount of activation involved in the process of comparison of two objects in a sentence (or word-pair), that is, the N400 varies according to the effort used to undertake comparisons that depend on how those two objects are distant in meaning one from another.

One of the first ERP studies that considered the Conceptual Mapping
Model was conducted by Coulson and Van Petten (2002

220
(Eg."The ring was made of tin, with a pebble instead of a gem") was thought to be an expression of an intermediate mapping effort in which the critical word of a sentence would near certain attributes of context similarly to metaphorical sentences (Eg."After giving it some thought, I realized the new idea was a gem").
Their results confirmed that hypothesis with N400 amplitudes varying across conditions being largest to metaphoric, intermediate for literal mapping and lowest to literal sentences. These results suggest that the higher the distance of two concepts the larger the N400 amplitude.
Other prominent studies have reached the same conclusions but using other experimental settings. Arzouan,  used word-pairs and found that the smallest N400 amplitude occurred in the literal condition and then was progressively larger in the following conditions respectively: conventional metaphor< novel metaphors < unrelated pairs. The authors explained their results in terms of effort rates applied to establish mapping between concepts and support that this is the main aspect modulating the N400 amplitudes. They argued that it is important to maintain the literal meanings for concept mapping and in a later step accessing the metaphorical meaning.
The Graded Salience Hypothesis is another possible explanation for this graded pattern found across studies. According to this model, the initial comprehension process is not really influenced by figurative content but by salience characteristics of sentences and content words. In this view, factors such as familiarity, conventionality, frequency and cloze probability facilitate the retrieval of word knowledge and consequently, sentence comprehension (GIORA, 1999(GIORA, , 2002. In accordance with this model, the N400 is mostly linked with the expectancy offered by the previous context on the critical word, that is, an unexpected word finalization would produce larger N400 amplitudes in comparison to more familiar sentences (salient sentences in this model refer to more familiar sentences) (DE GRAUWE et al., 2010;LAURENT et al., 2006;STRANDBURG et al., 1997). Also, it is possible to see these same patterns considering idioms' studies. The results of Laurent and colleagues (2006) suggest that salience of a sentence is a strong factor modulating the N400. The authors demonstrate in a well-controlled study that compared two categories of idioms (strong and weak idiomaticity) to literal expressions that N400 results followed the graded pattern, in which larger amplitudes were found for weak salience conditions (DE GRAUWE et al., 2010;GOLD-STEIN et al., 2012;LAURENT et al., 2006) In Access View has also received some support. This hypothesis proposes that figurative and literal meanings present the same course of processing as long as they share the same quantity of supportive context. It received some support by the results of Iakimova et. al. (2005), a clinical study, in healthy participants did not find significant differences in ERPs between figurative and literal sentences.
In that case, the major objective of this study was to analyses differences in processing between individuals with depression in comparison with healthy participants. In the present review it was very difficult to find studies about figurative processing that lent support to the direct view. There are two main theories about the types of processing indexed by the N400. In one view, the N400 is supposed to index a post-lexical process involved in semantic integration of the final word to the previous context. In another view, it is proposed that the N400 is a marker of lexical retrieval (from memory storage) (FEDERMEIER, 2014;KUTAS, 2006;LAI et al., 2009;LAU et al., 2008;NIEUWLAND;VAN BERKUM TZUYIN, 2006;CURRAN, 2013;WEILAND et al., 2014). Critically, it has also been proposed that both word meaning and world knowledge are recruited and integrated very rapidly by similar brain regions (HAGOORT; LEVIN-SON, 2004). These finding is particularly important as it suggests that what words and stimuli are used can possibly bias what affects the most the observed N400.
These two views seem similar or even the same phenomena, but they have particularities that may lead to different explanations for the process studied (FEDERMEIER, 2014;FEDERMEIER, WLOTKO;DE OCHOA-DEWALD;KUTAS, 2007;KUTAS, 2006).
Considering the mapping model, the effort for comprehension resides on the process of conceptual integration between items from a sentence. The as-

222
sumption that the N400 is a marker of semantic integration would point that the attribution of literal or figurative senses is a context integration process that happens in around 400 milliseconds, that is, in a relative early temporal window, considering that semantic integration is a conscious process.
Now, considering the N400 as a marker of lexical retrieval from memory, the mapping theory would produce a different explanation, indicating that around 400 milliseconds an unconscious process would occur: the recovery of word information that matches (or not) with previous context (even when context is just one word). In this view, semantic integration is characterized as late process which might be indexed by other ERP components (as P600, LPC, LPP).
In figurative language studies, this type of explanation was explicitly considered just by colleagues (2009, 2013) that discussed each possibility according with the functional role assumed for the N400.
Taking these functional views into consideration, the salience theory seems a better model for explaining the influence of linguistic factors on N400 amplitudes in the first window of processing. Salient meanings are accessed before than non-salient ones due to a memory facilitation process. Non-salient stimuli would need additional inferential processes best explained by later ERP components and in behavioral measures by larger RTs. In addition, considering the lexical-retrieval view of N400 there is a possibility of some overlap on the type of ERP pattern between the Graded Salience Hypothesis and Conceptual Mapping Views. It is very well established on the literature that lexical retrieval is largely influenced by factors such as cloze probability, familiarity, conventionality what fits with the idea that salience affects N400 amplitudes (FEDERMEIER, 2014;LAU et al., 2008). GIORA (2002) considers that the salience model represents aspects of an initial step of sentence processing. If the semantic integration view of N400 is correct, this would imply that lexical retrieval occurs in even earlier windows and On a well-controlled study Bambini et al. (2016) manipulated the context and found that in conditions of supportive context (see their experiment 2) there are no N400 differences between literal and figurative conditions. In their study, the N400 was greatly suppressed by context and differences on late ERP windows were found. On the other hand, the lack of context (experiment 1) produced a biphasic pattern in which N400 negativity is found followed by a later positive component such as P600 or LPC for metaphorical conditions. That biphasic pattern is recurrent across studies that support indirect views of figurative language, as those patterns of ERPs matches with the idea that the N400 is an index of lexical retrieval and that in later windows other processes would take place (i.e analysis and reanalysis of semantic aspects of sentence) (BAMBINI et al., 2016).
In agreement with this, results of studies with idiomatic expressions show that strongly lexicalized idioms (items with high cloze probability) don't elicit the typical N400 deflections seen in other types of figurative sentences. It is frequently argued that this finding reflects the process in which the entire phrase structure is accessed from memory. Thus, conditions that do not present lexical retrieval would not generate N400 deflections. In weak idiomaticity (low cloze probability) or literal conditions the presence of the N400 deflection is thought to reflect of the role of expectancy of the final sentence word disambiguating the previous context (LAURENT et al., 2006;VESPIGNANI et al., 2010).
Combined, these results also satisfy the proposals of a probabilistic factor affecting word retrieval in written or auditory comprehension processes (eg. reading or listening sentences). Our brains automatically pre-activate upcoming words according to the constraints provided by the previous context in a mechanism that facilitates reading or listening processes (DELONG; URBACH; KUTAS, 2005;FEDERMEIER, 2007;TROYER;KUTAS, 2020). If expectation is broken by an unexpected word (low cloze probability condition) the result is a larger N400 deflection that denotes search mechanisms (from memory) to retrieve the semantic characteristics of this new lexical item in the sentence being processed (DE-LONG et al., 2005;FEDERMEIER et al., 2007). In addition, the N400 amplitude can be modulated by the semantic relationship between the expected and the unexpected word. The N400 amplitudes are smaller when those items share many semantic features (eg. hammer vs saw that belong to the same semantic category). Thus, when the unexpected word is semantically related to the expected word a reduced N400 effect is usually explained in terms of a facilitation of the lexical retrieval process (WLOTKO; FEDERMEIER, 2015).
The studies sample selected for this review seems to be more supportive of an early automatic processing view of the N400 evoked-response. This support could evidence that figurative processing, if it is indexed by N400, might not comprise conscious processes or integrative difficult of semantic in the previous context.
The assumption that the N400 appears for unexpected words (is a marker of expectations breaches) offers a better explanation for the results seen in figurative sentences in which the N400 varies with some psycholinguistic aspects linked with the frequency and familiarity of some word in certain context (i.e great correlation of N400 with Cloze Probability)(KUTAS; FEDERMEIER, 2014).
Also from all these studies, just three discussed the implications of their findings according to each functionality view (BAMBINI et al., 2016;LAI et al., 2009;CURRAN, 2013 It is also important to highlight that many studies don't make this functional approach very clear. Specifically, studies that tried to characterize clinical subgroups or focused on the localization of the N400 did not provide explanations for the cognitive mechanisms that may be implied. In fact, in those studies, no functional view is chosen to explain results (COULSON, 2007;FORGÁCS et al., 2015;LORUSSO et al., 2015;PROVERBIO et al., 2009;SCHNEIDER et al., 2014SCHNEIDER et al., , 2015SOTILLO et al., 2005;STRANDBURG et al., 1997). Although this is not a problem in studies not aimed at comprehension of the time-course of cognitive processing, they may lead to incorrect conclusions. For example, if it is assumed that the N400 is indexing lexical retrieval, the absence or different modulation of this component in Schizophrenia would mean that pathology affects processes of lexical retrieval and not processes of semantic integration of stimuli to previous context.
The debate about figurative models is far to be finished. This review emphasizes the need of deeper analyses and interpretations of the N400 functionality to provide more precise insights of electrophysiological applications in figurative language research. In addition, we believe that figurative language research may bring evidence for the functionality of N400.
This can build up certain robust views about functionality of N400 if authors appropriate the functionality debate on experimental design or discuss the N400 in effect studied. Here we show that it is not all authors who consider this issue for results discussion and how this implicates in model explanations (LAI et al., 2009). It is undeniable the importance of N400 theory aspect for support certain figurative processing model.

Final Considerations
Our review aimed to show a general panorama and discussion of the use of N400