1 

Rafał Jończyk 

 

Affective language processing in a native 

and non-native mind: A neuropragmatic 

perspective 

Przetwarzanie języka afektywnego w języku 

ojczystym oraz obcym: Perspektywa 

neuropragmatyczna  

Praca doktorska 

na Wydziale Anglistyki 

Uniwersytetu im. Adama Mickiewicza w Poznaniu 

pod kierunkiem prof. Romana Kopytko oraz  

Dr Katarzyny Bromberek-Dyzman 



 2 

Acknowledgments 

First and foremost, I would like to thank my supervisors, Prof. Roman Kopytko and Dr. 

Katarzyna Bromberek-Dyzman for their incredible support, patience, and understanding. I 

am particularly grateful for them offering me a lot of freedom in my scientific endeavours, 

and being always there when I was in need of their expertise and advice. The mentoring 

they have given me deserves more than a few sentences atop a page in a dissertation. 

I am also eternally grateful to Prof. Guillaume Thierry from Bangor University, UK 

who took me under his wing during my stay in Bangor, opened his lab for me and offered 

guidance and mentoring in the exciting process of conducting my first electrophysiological 

experiments. Also, I would like to thank the whole BULET team, particularly Dr. Bastien 

Boutonnet for introducing me to electrophysiological data analysis in Matlab, being of 

great help and support during my stay in Bangor, and for constructive feedback on Chapter 

3. Thank you to Kamil Musiał and Katie Hoemann who finished the EEG data collection 

for experiment 2 on my behalf after my departure, and thus made it possible to bring the 

project successfully to an end. Last but not least, thank you to Vasco Bull, who offered a 

roof over my head during my first two weeks in Bangor, and became a good friend over the 

next 3 months.  

I would also like to thank Halszka Bąk for her friendship as well as insightful 

discussions. We set out on our PhD journeys together, which made me feel that I am not 

alone in the process. Also, a warm thank you to Prof. Aneta Pavlenko from Temple 

University, USA for giving me support and advice in this exciting journey, as well as for 

her constructive feedback on the written work. 

Finally, my PhD adventure would have not been the same had it not been for the 

love and support of those closest to me. I am forever in your debt.   

As a final remark, please note that part of the present dissertation has been already 

submitted for publication (experiment 2; Jończyk et al. forthcoming). 

 

    



 3 

Poznań, dnia ............................ 

 

 

OŚWIADCZENIE 

 

 

Ja, niżej podpisany Rafał Jończyk student Wydziału Anglistyki Uniwersytetu im. Adama 

Mickiewicza w Poznaniu oświadczam, że przedkładaną pracę dyplomową pt:  

„Affective language processing in a native and non-native mind: A neuropragmatic 

perspecitve” napisałem samodzielnie. Oznacza to, że przy pisaniu pracy, poza 

niezbędnymi konsultacjami, nie korzystałem z pomocy innych osób, a w szczególności nie 

zlecałem opracowania rozprawy lub jej części innym osobom, ani nie odpisywałem tej 

rozprawy lub jej części od innych osób. Oświadczam również, że egzemplarz pracy 

dyplomowej w formie wydruku komputerowego jest zgodny z egzemplarzem pracy 

dyplomowej w formie elektronicznej. 

Jednocześnie przyjmuję do wiadomości, że przypisanie sobie, w pracy dyplomowej, 

autorstwa istotnego fragmentu lub innych elementów cudzego utworu lub ustalenia 

naukowego stanowi podstawę stwierdzenia nieważności postępowania w sprawie nadania 

tytułu zawodowego. 

 

[TAK]* - wyrażam zgodę na udostępnianie mojej pracy w czytelni Archiwum UAM 

[TAK]* - wyrażam zgodę na udostępnianie mojej pracy w zakresie koniecznym do ochrony 

mojego prawa do autorstwa lub praw osób trzecich 

 

*Należy wpisać TAK w przypadku wyrażenia zgody na udostępnianie pracy w czytelni Archiwum UAM, 

NIE w przypadku braku zgody. Niewypełnienie pola oznacza brak zgody na udostępnianie pracy.  

 

   

 
 

(czytelny podpis studenta) 



 4 

Table of contents 

ACKNOWLEDGMENTS .................................................................................................... 2 

TABLE OF CONTENTS ..................................................................................................... 4 

LIST OF TABLES ................................................................................................................ 7 

LIST OF FIGURES .............................................................................................................. 8 

INTRODUCTION .............................................................................................................. 11 

CHAPTER 1 : AFFECT AND LANGUAGE ................................................................... 14 

1.1. INTRODUCTION ........................................................................................................... 14 

1.2. WHAT IS AN EMOTION?: PAST TO PRESENT ................................................................. 15 

1.2.1. Emotions as basic, natural kinds ........................................................................ 18 

1.2.1.1. Critical evaluation ........................................................................................ 19 

1.2.2. Emotions as appraisals ....................................................................................... 21 

1.2.2.1. Critical evaluation ........................................................................................ 23 

1.2.3. Emotions as psychological constructs ................................................................ 24 

1.2.3.1. Critical evaluation ........................................................................................ 26 

1.2.4. Does the concept of emotion serve any useful purpose in scientific psychology?

 ...................................................................................................................................... 27 

1.3. THEORETICAL FRAMEWORK ........................................................................................ 28 

1.3.1. Psychological construction: a framework for the present investigation ............ 28 

1.3.2. Core affect ........................................................................................................... 29 



 5 

1.3.3. Affective evaluations ........................................................................................... 32 

1.3.4. Affective primacy hypothesis .............................................................................. 34 

1.3.5. Unconscious affect .............................................................................................. 37 

1.3.6. The neuroscience of affect .................................................................................. 39 

1.4. AFFECTIVE PROCESSING OF VERBAL STIMULI .............................................................. 44 

1.4.1. Affect – language interface: A psycholinguistic perspective .............................. 45 

1.4.1.1. Processing advantage of affective words: Behavioural evidence ................ 45 

1.4.1.2. Processing advantage of affective words: Neuroimaging evidence ............ 47 

1.4.1.3. Effects of affective valence in sentences: Electrophysiological evidence ... 51 

1.4.2. Affect – language interface: A Pragmatic perspective ....................................... 53 

1.4.3. Affective Neuropragmatics ................................................................................. 58 

1.5. CONCLUSION ............................................................................................................... 60 

CHAPTER 2 : AFFECT PROCESSING IN THE NON-NATIVE LANGUAGE(S) ... 62 

2.1. INTRODUCTION ........................................................................................................... 62 

2.2. AFFECTIVE PROXIMITY OR AFFECTIVE DISTANCE? BILINGUALS IN PSYCHOTHERAPY.. 64 

2.3. AFFECT MANIFESTATION IN BILINGUALS’ AUTOBIOGRAPHIC MEMORY........................ 66 

2.4. BILINGUALS’ AFFECTIVE REPERTOIRES FROM AN INTROSPECTIVE PERSPECTIVE ......... 69 

2.4.1. The Bilingualism and Emotions Questionnaire (BEQ) ....................................... 70 

2.4.2. Bilinguals’ perception of affective intensity in their languages ......................... 70 

2.4.3. Bilinguals’ choice of language for affect manifestation ..................................... 72 

2.4.4. Summary ............................................................................................................. 73 

2.5. AFFECTIVE LANGUAGE PROCESSING IN COGNITIVE PARADIGMS .................................. 74 

2.5.1. Memory advantage for affective words .............................................................. 74 

2.5.2. Allocating attentional resources to affective words: Emotional Stroop and RVSP 

tasks .............................................................................................................................. 78 

2.5.3. Processing advantage for affective words: Affective priming and lexical 

decision paradigms ....................................................................................................... 81 

2.5.4. Summary ............................................................................................................. 83 

2.6. AFFECTIVE LANGUAGE PROCESSING IN PSYCHOPHYSIOLOGICAL PARADIGMS ............. 84 

2.7. AFFECTIVE LANGUAGE PROCESSING IN NEUROIMAGING PARADIGMS .......................... 88 



 6 

2.8. CONCLUSION ............................................................................................................... 92 

CHAPTER 3 : THE STUDY.............................................................................................. 94 

3.1. INTRODUCTION ........................................................................................................... 94 

3.2. EXPERIMENT 1 ............................................................................................................ 95 

3.2.1. Objectives and hypotheses .................................................................................. 95 

3.2.2. Materials and methods ........................................................................................ 96 

3.2.3. Results and discussion ...................................................................................... 103 

3.2.3.1. Behavioural data ........................................................................................ 103 

3.2.3.2. Electrophysiological data ........................................................................... 107 

3.3. EXPERIMENT 2 .......................................................................................................... 112 

3.3.1. Objectives and hypotheses ................................................................................ 112 

3.3.2. Materials and methods ...................................................................................... 114 

3.3.3. Results and discussion ...................................................................................... 117 

3.3.3.1. Behavioural data ........................................................................................ 117 

3.3.3.2. Electrophysiological data ........................................................................... 121 

3.4. DISCUSSION .............................................................................................................. 127 

CHAPTER 4 :GENERAL DISCUSSION AND SUMMARY ...................................... 128 

LIMITATIONS.................................................................................................................... 138 

IMPLICATIONS FOR FUTURE RESEARCH ............................................................................ 138 

CONCLUSION ................................................................................................................... 139 

ABSTRACT ....................................................................................................................... 140 

STRESZCZENIE .............................................................................................................. 142 

REFERENCES .................................................................................................................. 145 

APPENDICES ................................................................................................................... 179 

APPENDIX 1 ..................................................................................................................... 179 

APPENDIX 2 ..................................................................................................................... 187 



 7 

List of tables 

Table 1. Sociobiographical and linguistic information about monolingual and bilingual 

participants. The measures provided correspond to means. The measures provided in 

brackets reflect Standard Error of the Mean (SEM). ..................................................... 97 

Table 2. Exemplary noun-adjective dyads. ........................................................................... 98 

Table 3. Stimuli characteristics for prime nouns. ................................................................. 99 

Table 4. Stimuli characteristics for target adjectives. ........................................................... 99 

Table 5. Post-experimental ratings of prime and target valence. ........................................ 100 

Table 6. Mean number of accepted epochs per condition in monolingual and bilingual 

participants in experiment 1. ........................................................................................ 101 

Table 7. Mean number of accepted epochs per condition in monolingual and bilingual 

participants in experiment 2 ......................................................................................... 115 

Table 8. A complete set of noun-adjective dyads from experiment 1. “-“ depicts negative 

valence of a prime or target; “+” depicts positive valence of a prime or target; “0” 

depicts neutral valence of a prime; “*” indicates an unrelated noun-adjective dyad. .. 179 

Table 9. A complete set of sentences from experiment 2. “-“ depicts negative valence of 

sentence context or target; “+” depicts positive valence of sentence context or target; 

“0” depicts neutral valence of sentence context; “*” indicates a semantically 

incongruent sentence. ................................................................................................... 187 



 8 

List of figures 

Figure 1. A representation of core affect in a two-dimensional circumplex model (after 

Russell 2012: 86)............................................................................................................ 30 

Figure 2. The visualization of the 7 brain networks (after Yeo et al. 2011: 1137), with an 

added description of the regions included in each of the network (see Yeo et al. 2011; 

Oosterwijk et al. 2012) ................................................................................................... 42 

Figure 3. Mean reaction times (bars, left axis) and error rates (bullets, right axis) for 

English monolingual and Polish-English bilingual participants in experiment 1 in the 

following conditions: related negative (R(-)), related positive (R(+)), unrelated negative 

(U(-)), unrelated positive (U(+)). Error bars reflect SEM. ........................................... 107 

Figure 4. EPN elicited by target adjectives in English monolinguals. A: Waveforms 

illustrate brain potential variations computed via linear derivation from 19 parieto-

occipital electrodes (O1, O2, OZ, PZ, P1, P2, P3, P4, P5, P6, P7, P8, POZ, PO3, PO4, 

PO7, PO8, PO9, PO10). Shaded areas represent significant difference between 

conditions in the 200 – 300 ms time window. B: EPN at selected 6 electrode sites 

where the effect was maximal. The schematic head reflects difference topography of 

cortical responses to positive minus negative adjectives 200 - 300 ms post-adjective 

onset at the electrode sites of interest. .......................................................................... 108 

Figure 5. N400 elicited by related and unrelated target adjectives in Polish-English 

bilinguals and English monolinguals. All waveforms represent brain potential 

variations computed via linear derivation from 10 central electrodes (FZ, FC1, FC2, 

FCZ, C1, C2, CZ, CP1, CP2, CPZ). Shaded areas represent significant difference 



 9 

between conditions in the 350 – 500 ms time window. The schematic head reflects 

difference topography of cortical responses to related minus unrelated noun-adjective 

dyads 350 – 500 ms post-adjective onset at the electrode sites of interest. ................. 110 

Figure 6. N400 and LPC elicited by target adjectives in Polish-English bilinguals. A: 

Waveforms illustrate brain potential variations computed via linear derivation from 10 

centro-frontal electrodes (FZ, FC1, FC2, FCZ, C1, C2, CZ, CP1, CP2, CPZ). Shaded 

areas represent significant difference between conditions in the 350 – 500 and 600 – 

800 time window. B: N400 and LPC at selected 7 electrode sites where the effects were 

maximal. The schematic heads reflect difference topography of cortical responses to 

positive minus negative target adjectives 350 – 500 ms and 600 – 800 ms post-

adjective onset at the electrode sites of interest. .......................................................... 111 

Figure 7. Mean reaction times (bars, left axis) and error rates (bullets, right axis) for 

English monolingual and Polish-English bilingual participants in experiment 2 in the 

following conditions: congruent negative (C(-)), congruent positive (C(+)), incongruent 

negative (I(-)), incongruent positive (I(+)). .................................................................. 121 

Figure 8. N400 elicited by congruent and incongruent target adjectives in Polish-English 

bilinguals and English monolinguals. All waveforms represent brain potential 

variations computed via linear derivation from 10 centro-frontal electrodes (FZ, FC1, 

FC2, FCZ, C1, C2, CZ, CP1, CP2, CPZ). Shaded areas represent significant difference 

between conditions in the 280 – 550 ms time window. The schematic heads reflect 

difference topography of cortical responses to congruent minus incongruent sentence-

final adjectives 280 – 550 ms post-adjective onset at the electrode sites of interest. .. 123 

Figure 9. N400 and LPC elicited by target adjectives in Polish-English bilinguals. A: 

Waveforms illustrate brain potential variations computed via linear derivation from 10 

centro-frontal electrodes (FZ, FC1, FC2, FCZ, C1, C2, CZ, CP1, CP2, CPZ). Shaded 

areas represent significant difference between conditions in the 280 – 550 and 600 – 

800 time window. B: N400 at selected 7 electrode sites where the effect was maximal. 

The schematic head reflects difference topography of cortical responses to English 

negative minus Polish negative target adjectives 280 – 550 ms post-adjective onset at 

the electrode sites of interest. ....................................................................................... 124 



 10 

Figure 10. C: N400 at selected 7 electrode sites where the effect was maximal. The 

schematic head reflects difference topography of cortical responses to English positive 

minus Polish positive target adjectives 280 – 550 ms post-adjective onset at the 

electrode sites of interest. ............................................................................................. 125 

Figure 11. LPC at selected 7 electrode sites where the effect was maximal. The schematic 

head reflects difference topography of cortical responses to Polish positive minus 

Polish negative target adjectives 600 – 800 ms post-adjective onset at the electrode 

sites of interest. ............................................................................................................. 126 

 



 11 

Introduction 

Communication is an arena of expression, exchange, and sharing of people’s thoughts, 

opinions, beliefs, feelings, or attitudes. Two essential, interacting and integral expressive 

agents in communication are affect (e.g. emotions, feelings, attitudes, beliefs) and language 

(its verbal and non-verbal manifestation). The relationship between affect and language is 

highly reciprocal. Affect finds an outlet on all levels of non-verbal (e.g. gestures, body 

posture, facial expression) as well as verbal (e.g. syntax, semantics, pragmatics) language. 

In turn, language may have the capacity to modify people’s affective states (e.g. by means 

of swear words, compliments, or poetry). 

In today’s world, individuals’ communicative interactions are oftentimes 

constructed in and coloured by a multilingual context. Indeed, it has been estimated that 

more than half of the world population speaks more than one language (Grosjean 1984, 

2010). Therefore, this multilingual ingredient should be taken into consideration when 

analysing communicative interactions, and thus constitute an important factor in the 

analysis of affective manifestations in communication. In fact, research on affective 

language in bilingualism has shown that bilingual individuals often report substantial 

differences in the way they perceive and express affective information in their first and 

second language, with the second language being more affectively detached than the first 

(see Dewaele 2010; Pavlenko 2005). In the search for behavioural and psychophysiological 

correlates of this phenomenon, however, psycholinguistic and neuroimaging paradigms 

reported essentially no measurable differences between affective word processing in the 

native and non-native language (see Pavlenko 2012). What seems to have been put in the 



 12 

shade in these paradigms, however, is a pragmatic perspective that through the 

implementation of natural linguistic context might provide a more comprehensive and 

authentic picture of bilinguals’ affective repertoires in communicative encounters. In the 

present dissertation I will therefore adopt a pragmatic approach to the investigation of 

affective language in bilingual speakers that might account for the inconsistencies reported 

in behavioural, psychophysiological and neuroimaging studies. Specifically, I set out to 

investigate how the build-up of contextual information might modulate behavioural and 

electrophysiological responses to affective words in the native and non-native language in 

two experiments on immersed Polish-English bilinguals. Hence, this study aims to explore 

the available psycho- and neuro-linguistic findings in the research on affective language 

processing in bilingualism and, owing to the “pragmatic twist” in the design, provide a 

more reliable picture of the phenomenon under investigation. 

Chapter 1 will discuss the theoretical and empirical background of the relationship 

between affect and language. It will present a historical perspective on the concept of 

emotion, demonstrating how its understanding has been shaped over the past century. Here, 

the mainstream models of emotion will be discussed along with their critical evaluation. 

Particular attention will be devoted to the premises of the psychological construction model 

(Barrett and Russell 2015) and the concept of core affect that provide the theoretical 

foundation for the present investigation. Finally, the interaction between affect and 

language will be discussed from a psycholinguistic and pragmatic perspective. It will be 

brought to the fore that in order to better understand the mechanisms governing the 

manifestation of affective language in communication, there is a need for an introduction of 

linguistic context into an experimental design, likewise to experiments in the field of 

neuropragmatics. 

Chapter 2 will build on the discussion on affect – language interface from Chapter 

1, but – importantly – it will extend it to a bilingual context. It will provide a critical review 

of a wide array of studies investigating affective repertoires of bilingual speakers in such 

contexts as psychotherapy or psycholinguistic laboratory, and using such measures as self-

reports or electroencephalography. As it will be shown, at present there is a marked 

discrepancy between the findings in the field. While clinical and introspective studies 



 13 

suggest that the second language may be affectively more distant than the first language, 

cognitive, psychophysiological and neuroimaging evidence report essentially no 

measurable differences in affect processing in L1 and L2. As in Chapter 1, it will be argued 

that such inconsistency in bilingual research on affective language might stem from 

unnatural experimental designs and the absence of linguistic context in psycho- and neuro-

linguistic paradigms.  

Chapter 3 will describe two experiments whose aim is to directly address the 

aforementioned limitation of psycho- and neuro-linguistic studies on affective language in 

bilingual speakers, i.e. the absence of linguistic context in experimental design. The 

described experiments constitute the first attempt to directly investigate the impact of a 

build-up of contextual information on the processing of the same set of affective words in 

the first and second language of bilingual individuals. Also, both behavioural and 

electrophysiological data were collected in experiment 1 and 2 to obtain a holistic view on 

the phenomenon under investigation. As it will be reported and discussed, the collected 

data provides compelling evidence of the influence of natural sentence context on 

bilinguals’ affective responses to stimuli. Specifically, the present study demonstrates a 

differential modulation of electrophysiological responses to affective sentences, but not 

single words, in L1 and L2. This finding offers the first neurocognitive interpretation of 

findings reported in clinical and introspective studies.  



 14 

Chapter 1: Affect and language 

1.1. Introduction 

The aim of this chapter is two-fold. First, its goal is to bring to the fore the 

reciprocal relationship between affect and language in communication, and to connect these 

two pieces into a coherent whole, consistent with affective pragmatics – an emerging field 

of theoretical and empirical research whose aim is to look into the complex affect-language 

interface in communicative interactions. Second, its goal is to go beyond the theoretical 

discussion of affect-language interface in communication and to put it into an empirical, 

neuroscientific perspective; a perspective I will refer to as affective neuropragmatics. 

Structure-wise, the first chapter is divided into three main sections whose aim is to 

gradually build the foundation for understanding the principles of affective 

neuropragmatics.  

In the first section I review the past and present approaches to emotion, highlighting 

the ambiguities and controversies around this concept. This section provides the motivation 

behind a recent theoretical and empirical re-focus on the analysis of more general affective, 

not emotional, phenomena.  

The second section is devoted to affective phenomena. I begin with a discussion of 

core affect, an essential substrate of any affective phenomenon, and a new framework in 

affective sciences. I follow with a discussion of the everyday manifestation of affect in the 

domain of evaluation and the hypothesis about the primacy of affect over cognition. 



 15 

Finally, I address the issue of unconscious affect and provide a review of studies on the 

neural underpinnings of affective phenomena. 

The third and final section of Chapter 1 is reserved for the discussion of how affect 

manifests itself in verbal language. Here, I review the available evidence from two 

perspectives – a psycholinguistic perspective and a pragmatic perspective – and point out 

their weak points: the focus on examining decontextualized affective stimuli 

(psycholinguistics) and the focus on the rational and cognitive aspects of a communicative 

interaction, leaving behind its affective substrate (traditional, cognitive pragmatics). For a 

solution, I turn to affective neuropragmatics that assembles what is probably best in the two 

approaches – the methodological precision and tools (psycholinguistics) with the focus on 

contextual effects in communication (cognitive pragmatics) - as well as addresses and 

resolves their above-mentioned weak points. As such, affective neuropragmatics, by its 

focus on neurophysiological measures of affective language in communication, constitutes 

the methodological framework for the present investigation.  

Throughout, I will use the term emotion or emotional to refer exclusively to the 

common sense emotion categories, as delineated by basic emotion theorists (section 1.2.1;  

anger, happiness, disgust) and reflected in subjective experience or language (e.g. he is 

happy). The term affect or affective will be used as a general umbrella term for anything 

emotional (e.g. emotion, mood, motivation, preference/liking, attitude, valence, arousal). 

Core affect, in turn, will be used in its original sense, as delineated by psychology 

construction theorists (section 1.2.3; e.g., Russell and Barrett 1999; Russell 2003, 2009, 

2012, 2015; Barrett 2006a), to refer to the very elemental affective substrate, a basic 

neurophysiological experience that feels ‘good’ or ‘bad’.  

1.2. What is an emotion?
1
: Past to Present 

More than any other species, we are beneficiaries and victims of a wealth of emotional 

experience. (Dolan 2002: 1191) 

                                                 
1
The title of this section refers to the seminal work by William James (1884) with the same title. 



 16 

Despite over a century of theory and research, the understanding of what emotion is and 

how it should be defined remains unclear. According to Mulligan and Scherer (2012), 

“[t]here is no commonly agreed-upon definition of emotion in any of the disciplines that 

study this phenomenon” (2012: 345). Many scholars argue that this confusion about the 

definition of emotion has significantly delayed progress in both theoretical and empirical 

understanding of the phenomenon (see Walla and Panksepp 2013; LeDoux 2014, 2012b, 

2015; Dixon 2012; Russell 1991; Kleinginna and Kleinginna 1981; Russell 2012; Russell 

and Barrett 1999; Panksepp and Watt 2011; Duffy 1934a, 1934b). In this subsection, my 

aim is to present a brief history of the emotion concept as well as to discuss three main 

currents of thought about the nature of emotion and provide their critical evaluation.  

The spark of interest in contemporary emotion research was ignited by Charles 

Darwin’s (1809 – 1882) publication of The Expression of the Emotions in Man and 

Animals (1872), in which Darwin expressed his view that emotions are ‘states of mind’ that 

trigger stereotypic patterns of behaviour. Darwin’s contemporary, William James (1842 – 

1910), disagreed with this observation and postulated that emotion is caused by visceral 

changes in the body. In his seminal essay entitled What is an emotion? James (1884) 

writes: 

Our natural way of thinking about these standard emotions [currently referred to as basic 

emotions, e.g. fear, anger] is that the mental perception of some fact excites the mental 

affection called the emotion, and that this latter state of mind gives rise to the bodily 

expression. My thesis on the contrary is that the bodily changes follow directly the 

PERCEPTION of the exciting fact, and that our feeling of the same changes as they occur IS the 

emotion [emphasis in the original, RJ]. (1884: 189–190) 

Therefore, to James (1884, 1913) emotional states were generated as a result of 

physiological reactions in the body, not the other way around. This view was in turn 

criticized by Walter Cannon (1871 – 1945), an American physiologist, who, together with 

his student Philip Bard (1989 – 1977), formulated an opposing model of emotion referred 

to as the Cannon-Bard theory (Cannon 1927). Based on extensive research of the 

autonomic nervous system (ANS) in a cat, Cannon concluded that emotions might be 

elicited in the absence of autonomic feedback. Furthermore, Cannon claimed that feelings 

and visceral reactions were separate, independent processes engendered by the brain in the 



 17 

course of processing of emotional stimuli. These views were articulated in Cannon’s 

seminal work entitled The James-Lange theory of emotions: A critical examination and an 

alternative theory (1927). For a detailed discussion of the two opposing theories by James 

(1884, 1890) and Cannon (1927), refer to Friedman (2010).  

The theoretical views on the nature of emotions expressed by Charles Darwin and 

William James have had an immense impact on emotion inquiry and research, and, among 

others, have provided the foundation for three main contemporary approaches to emotions. 

In the Darwinian tradition, emotions have been interpreted as being biologically given 

‘natural kinds’ that are hard-wired into the brain; they are observable in nature and 

recognized by the mind; they are complex, automatic reflexes elicited by stimuli in the 

environment (Ekman et al. 1987; Ekman 1992, 1993; Izard 1994, 2009; Tomkins 1962, 

1963; Panksepp 1998; Panksepp and Watt 2011; Arnold 1960). In the Jamesian tradition, 

emotions are considered ‘mental constructs’ that are elicited by more basic psychological 

processes; they are not recognized but constructed by the mind (Barrett 2006b, 2006a; 

Barrett and Bliss-Moreau 2009; Lindquist and Barrett 2012; Barrett 2013, 2011; Gendron 

and Barrett 2009; Russell 2012, 1994, 2003; Mesquita and Boiger 2014; Wundt 1902; 

James 1884, 1913; Schachter and Singer 1962). The followers of Darwin’s arguments have 

represented what is referred to as the natural kind (or basic emotions) view while the 

Jamesian tradition is thought to have inspired the appraisal approach to emotion (Arnold 

1960; Lazarus 1966, 1991; Frijda 1986, 1993, 2013; Scherer 1984; Ellsworth and Scherer 

2003; Scherer 2009; Mulligan and Scherer 2012), and laid the foundation for the 

psychological construction view of emotion (Russell 2009; Barrett 2011, 2013; Barrett and 

Russell eds. 2015). 

These approaches will be discussed in turn in the following subsections. It should be 

noted, however, that a detailed description of their theoretical and empirical bases is 

beyond the scope of this dissertation. For a comprehensive review of the historical 

development of emotion research and the presentation of the emotion approaches, refer to 

Gendron and Barrett (2009) and Strongman (2003). 



 18 

1.2.1. Emotions as basic, natural kinds 

The natural kind or basic emotions paradigm has guided contemporary research on emotion 

in the last three decades (LeDoux 2012b). This theory proposes a set of biologically given 

emotion categories (e.g. fear, anger), each being characterized by distinct neural and 

psychophysiological correlates; each thought to be the basic, primitive, universal; each 

thought to be the natural kind (Tomkins 1962, 1963; Ekman et al. 1987; Ekman 1992, 

1993; Izard 1993, 1994, 2009; Panksepp 1998; Panksepp and Watt 2011). In the natural 

kind view, emotions could be therefore compared to atoms such that they are thought to 

constitute the fundamental features of the mind and brain (Barrett 2011).  

The research on basic emotions and the idea of universality and essentiality of a 

number of emotion categories was based on empirical investigation of the perception of 

facial expressions among literate and pre-literate cultures (Ekman et al. 1969; Ekman and 

Friesen 1971; Izard 1971; Ekman and Friesen 1977; Ekman et al. 1987). The initial 

accounts of the basic emotions theory were presented in the paper by Ekman, Sorenson, 

and Friesen (1969) in which the researchers reported cross-cultural agreement in the 

recognition of photographs of facial expressions depicting happiness, fear, disgust-

contempt, anger, surprise, and sadness. These results were further supported by a 

simultaneous cross-cultural study by Izard (1969). Not long after the studies had been 

published did the classification of happiness, fear, disgust, anger, surprise, and sadness as 

the basic and universal emotional categories enter the psychology course books. The basic 

emotions model has become the standard view according to which emotional phenomena 

have been explained and investigated for the next 30 years (Russell 1991, 1994; LeDoux 

2012b). Many scholars seemed to acknowledge the basic emotion theory as a scientific law; 

for example, Matsumoto (1990: 195) argued that “the universality of facial expression of 

emotion is no longer debated in psychology”; Izard and Saxton (1988: 651–652) agreed 

that “[t]he evidence for the innateness and universality of the expressions of the 

fundamental emotions is sufficiently robust to consider Darwin’s hypothesis as an 

established axiom of behavioral science”; finally, Brown (1991: 26) affirmed that “[t]he 

conclusion seems inescapable: There are universal emotional expression”. 



 19 

1.2.1.1. Critical evaluation 

This “established axiom”, as argued by Izard and Saxton (1988: 652), however, did not 

escape criticism. Indeed, throughout the years the basic emotions theorists have been 

criticized for not reaching consensus on which emotions should be classified as basic and 

what the guidelines are for such a classification (LeDoux 2012b; Ortony and Turner 1990). 

Some of the basic emotion theorists introduced changes to their own sets of basic emotion 

categories. For example, Ekman (1992) extended his original set of 6 basic emotions with 

additional categories of awe, amusement, contempt, embarrassment, excitement, guilt, 

interest and shame. In a later paper, Ekman (1999) added some positive basic emotion 

categories such as amusement, pride in achievement, satisfaction, relief and contentment. 

At the same time, however, Ekman’s set of basic emotion categories has significantly 

differed from those proposed by other basic emotions theorists (Panksepp 1998, 2005; 

Panksepp and Watt 2011; Walla and Panksepp 2013; Izard 1994, 2009; Frijda 1986, 2013). 

Due to this confusion around the classification of basic emotion categories, some 

researchers questioned the validity of the basic emotion model (see Ortony and Turner 

1990; LeDoux 2012b). 

Further criticism of the theory has been directed at poor reliability of the forced-

choice paradigm implemented in typical experiments on emotion perception (for 

argumentation see Russell 1991, 1993, 1994; Walla and Panksepp 2013; Clore and Ortony 

2013; Ortony and Turner 1990; LeDoux 2012b; Gendron et al. 2012). In a forced-choice 

task, participants view posed expressions of emotions and are asked to select from a list of 

emotion words one that best describes the facial expression. Oftentimes, as was the case in 

the seminal work by Ekman et al. (1969), the list of choices fully corresponded to the 

instances of basic emotions expressed by the facial stimuli; filler items were rarely made 

available to participants. Contemporary research shows, however, that the forced-choice 

paradigm might skew the data to support the formulated hypothesis (Russell 1993; Widen 

et al. 2011; Lindquist et al. 2006; Gendron et al. 2012). For example, a recent study by 

Widen et al. (2011) found that in a free-labelling task, in which participants spontaneously 

labelled four facial expressions (contempt, shame, embarrassment, compassion), more than 



 20 

80% of attributions were made incorrectly. By contrast, in a forced-choice response format 

the accuracy ratings significantly improved (Widen et al. 2011). This study supports the 

hypothesis that perception of emotion is to a significant extent contingent upon the 

accessibility of linguistic cues (see the language-as-context hypothesis; Lindquist et al. 

2006; Gendron et al. 2012) such that when participants are asked to make a perceptual 

judgment on a facial expression in the absence of linguistic labels, their performance 

significantly deteriorates. These findings have challenged the reliability of the body of 

evidence supporting the universality of emotion perception that has provided the foundation 

for basic emotions theory. 

Another methodological concern expressed by some opponents of the basic emotion 

theory involves the ecological validity of the facial expressions used in standard emotion 

perception studies. According to Clore and Ortony (2013: 338), facial expressions – 

considered the “gold standard” for emotion differentiation – display posed emotions that 

are not typically encountered in social interactions. In a similar vein, Walla and Panksepp 

(2013) argue that facial expressions may not reliably represent human emotional experience 

in the first place:  

[A] facial expression is an emotion of the person the face belongs to. The image of a facial 

expression is not necessarily in itself a matching affective stimulus such as the scene that 

elicited affect in the person demonstrating the facial expression (…). It instantly becomes 

clear that indirect affective information as communicated via facial expression can be 

misinterpreted and actually lead to different affective processing and a different emotion in 

the observer of a facial expression. (Walla and Panksepp 2013: 99; see Sabatinelli et al. 2011) 

Finally, recent evidence from the field of human and animal neuroscience has 

questioned the basic emotions theory premise about the existence of hard-wired basic 

emotion circuits in the brain (e.g. LeDoux 2012b, 2014; Lindquist and Barrett 2012; 

Oosterwijk et al. 2012; Barrett and Satpute 2013; Lindquist et al. 2012; Kober et al. 2008; 

Duncan and Barrett 2007; Wager et al. 2008). Considered a leading basic emotions 

neuroscientist, Panksepp (1998, 2005, 2011; Panksepp and Watt 2011; Walla and Panksepp 

2013), based on extensive research, has identified 7 basic emotional circuits in the animal 

brain: SEEKING, RAGE, FEAR, LUST, CARE, PANIC, PLAY. It remains questionable, 



 21 

however, how the classification of basic, universal emotions proposed by Panksepp (see 

1998) correlates with that of Ekman’s (1992, 1999; Ekman et al. 1969).  

 Despite the critical accounts of the basic emotions theory, it remains to be said that 

this approach has not only provoked a lot of questions but in a way provided the foundation 

for the present investigation of emotion. Notably, it has also left a lasting legacy in the form 

of the development of the Facial Action Coding System (FACS; Ekman and Friesen 1977) 

that enables a comprehensive analysis of human facial movements. Finally, one of the key 

assumptions of the basic emotions theory – that there are innate emotion circuits in the 

brain – cannot be easily disclaimed due to the limited resolution of the available 

neuroimaging techniques (LeDoux 2012b). 

1.2.2. Emotions as appraisals 

The appraisal theory postulates that emotions are immediate and automatic responses to 

evaluations (or appraisals) and interpretations of the environment (Arnold 1960; Smith and 

Ellsworth 1985; Ellsworth and Scherer 2003; Lazarus 1991; Leventhal 1984; Ortony et al. 

1990; Clore and Ortony 2008, 2013; Scherer 1984; Mulligan and Scherer 2012). The 

appraisal theory is said to have been pioneered by Magda Arnold (1960) and Richard 

Lazarus (1966) and the roots of which trace back to the ideas expressed by James (1890) as 

well as to ancient philosophers’ reflections about emotions (Moors et al. 2013). The basic 

assumption of the theory, as originally postulated by Arnold (1960), is that humans 

constantly and implicitly evaluate the stimuli and events in their environment for personal 

relevance, and that such meaning analysis triggers an emotion. In a recent reformulation, 

appraisal theorists conceive of emotions as “adaptive responses which reflect appraisals of 

features of the environment that are significant for the organism’s well-being” (Moors et al. 

2013: 119). As such, emotions are conceptualized not as states but as dynamic processes. 

Depending on the context (personal, situational, cultural), the meaning analysis may vary, 

leading to an emergence of different emotions. Appraisal theories may therefore account 



 22 

for variability in emotion experience in interpersonal and cross-cultural contexts (Roseman 

1991; Moors et al. 2013; Moors 2014).  

Appraisal theories postulate that emotional events result from alterations in the 

underlying mechanisms or components. These components include appraisal (evaluation 

and interpretation of the environment), motivation (action tendencies), physiological 

responses, expressive behaviour, and subjective experience or feelings (see Moors et al. 

2013: 119–120; Moors 2014). The components are meant to be highly interactive such that 

alterations in a given component impinge on other components. Notably, appraisal – as an 

ingredient of an emotional event – has been also incorporated in other emotion theories 

(e.g. Ekman 1994; Russell 2003). What makes appraisal theories stand out is the fact that 

here appraisal lies at the core of emotion and constitutes the key component of an 

emotional event (Frijda 2013; Lazarus 1991; Ellsworth and Scherer 2003; Scherer 2009; 

Mulligan and Scherer 2012; Roseman 1991; Clore and Ortony 2008). Indeed, appraisal 

theorists have proposed a set of appraisal criteria to account for various types of emotional 

episodes that arise in stimulus-environment interactions, e.g. stimulus novelty, valence, or 

relevance for an individual’s goals (Brosch 2013; Moors 2014). 

As in the case of basic emotion theory, however, appraisal theory is not a 

homogenous model. Appraisal theorists differ, among others, with regard to the delineation 

of the set of the aforementioned appraisal criteria or features, their degree of automaticity 

or whether or not the appraisal features are processed in a fixed sequence (for a discussion 

see Moors et al. 2013; Moors 2014; Clore and Ortony 2008, 2013). Some authors (see 

Moors 2014; Barrett 2011) thus suggested that contemporary appraisal theories can be 

divided into two distinct strands (Moors 2014: 304). The first strand is reminiscent of the 

basic emotion theory whereby it focuses on the analysis of a limited set of causal 

antecedents of distinct mental events that are observable in nature and correspond to a 

specific set of emotion words such as ‘fear’ or ‘anger’ (Arnold 1960; Scherer 1984; Lazarus 

1991; Roseman 1991). In line with this view, appraisals cause emotions (Barrett 2011); or 

more specifically, “different patterns of appraisal elicit different emotions” (Roseman 1991: 

162). By contrast, the second strand of appraisal theories analyses emotional episodes in 

terms of the more general, underlying causal mechanisms (components), without the 



 23 

precondition that such processes have to be emotional in nature (Scherer 2009; Ortony et al. 

1990; Smith and Ellsworth 1985; Schachter and Singer 1962; Clore and Ortony 2008, 

2013). Here, appraisals do not cause but constitute emotions as psychological phenomena 

that arise from non-emotional ingredients (Moors et al. 2013; Moors 2014; Barrett 2011; 

Clore and Ortony 2008). Hence, the second strand of appraisal theories is by some scholars 

viewed as a constructionist approach (e.g. Barrett 2011), the discussion of which I defer to 

the following subsection. Some authors, however, remain sceptical about making such 

categorizations. For example, while Brosch (2013) acknowledges the many commonalities 

between appraisal and constructionist theories of emotion, he does not subsume appraisal 

theory under constructionist theories. Furthermore, Brosch (2013) seems to more explicitly 

demarcate the line between the non-modular appraisal and modular basic emotion theories, 

to quote:  

While some theorists have indeed developed appraisal profiles with the aim of specifying the 

elicitation of basic emotions (Roseman 1991), most appraisal theorists see emotional episodes 

as an ongoing emergent process that is characterized by continuous changes in the underlying 

appraisals, and focus on the dynamic nature of an emotional response (Frijda, 1986; Scherer 

& Ellsworth, 2009). Thus, most appraisal theorists would agree that there are as many 

different emotional states as there are different dynamic appraisal outcomes. (Brosch 2013: 

370) 

1.2.2.1. Critical evaluation 

Criticism of the appraisal theory has been mainly directed at two of its assumptions: a) the 

conception that appraisal is a causal mechanism, and b) the (mis)conception that appraisal 

is a controlled, cognitive process.  

The conception of appraisal as a process that triggers emotion has provoked 

considerable criticism in emotion research (e.g. Barrett 2012; Barrett et al. 2007; see Moors 

2013), also on the part of appraisal theorists (for a discussion see Moors 2014). For 

example, Clore and Ortony (2013), the advocates of the Ortony, Clore, Collins (OCC) 

appraisal model (see Ortony et al. 1990; Clore and Ortony 2008, 2013) suggest that 

“appraisals [are] characterizations rather than causes of emotions (…). Emotion may thus 



 24 

be constructed rather than triggered [emphasis mine, RJ]” (Clore and Ortony 2013: 9), thus 

comparing appraisal to a “sculptor” of emotional episodes (Clore and Ortony 2013: 9). In a 

similar vein, Barrett et al. (2007) have objected to such “functionalist assumptions that 

reduce the experience of emotion to its immediate causal relations” (Barrett et al. 2007: 

375).  

The second line of criticism concerns the cognitive, supposedly non-automatic 

nature of the appraisal process and thus questions the possibility of the influence of 

appraisal on the elicitation of rapid emotional responses (for a discussion, see Clore and 

Ortony 2008, 2013; Brosch 2013; Moors et al. 2013). Many appraisal theorists, however, 

highlight that appraisal may occur at both conscious and unconscious, automatic level (van 

Reekum and Scherer 1997; Leventhal and Scherer 1987; Mulligan and Scherer 2012; Frijda 

2013; Moors et al. 2013; Moors 2010; Brosch 2013). According to Frijda (2013: 106), 

“appraisal processes are, in principle, nonconscious. Their outcomes may be conscious, in 

how one sees and experiences emotional events (…)”. Frijda’s (2013) argument echoes 

Arnold’s (1960) original formulation of appraisals as “sense judgments” that are “direct, 

immediate, non-reflective, nonintellectual and automatic [in] nature” (Arnold 1960: 175). 

Precisely in which circumstances appraisal is an automatic/non-automatic phenomenon 

remains an empirical question. 

1.2.3. Emotions as psychological constructs 

With roots in the ideas of William James (1884, 1913), the psychological construction 

model considers emotional experiences as highly variable mental states constructed by 

basic, global processes (psychological primitives) that are not specific to emotion (Barrett 

2011, 2012, 2013; Gendron and Barrett 2009; Lindquist 2013; Russell 2003, 2009). While 

the first articulation of the psychological construction view on emotion is dated back to 

James’ essay in Mind entitled “What is an emotion?” (1884), the author dedicated more 

space for the discussion of emotion in “The Principles of Psychology” (1890): 



 25 

The trouble with the emotions in psychology is that they are regarded too much as absolutely 

individual things. So long as they are set down as so many eternal and sacred psychic entities, 

like the old immutable species in natural history, so long all that can be done with them is 

reverently to catalogue their separate characters, points, and effects. But if we regard them as 

products of more general causes (as 'species' are now regarded as products of heredity and 

variation), the mere distinguishing and cataloguing becomes of subsidiary importance 

[emphasis in the original, RJ]. Having the goose which lays the golden eggs, the description 

of each egg already laid is a minor matter. (James 1913: 449) 

James (1913) was also sceptical about the idea to introduce emotion as a new entity or 

discipline in light of evidence from other established disciplines that could already explain 

this phenomenon: “emotion is the resultant of a sum of elements, and each element is 

caused by a physiological process of a sort already well known” (James 1913: 453). Such 

views have been typically associated with a constructionist approach, expressed and 

reemphasized throughout the years by other researchers (e.g. Duffy 1934a, 1934b; LeDoux 

2012b, 2012a). For this reason I dedicate a separate subsection to specifically address the 

question whether the concept of emotion is necessary for further progress in the field (see 

section 1.2.4). 

At present, the psychological construction model of emotion is a family of different 

accounts (Russell 1980, 2003, 2009; Barrett 2014; Barrett et al. 2015; Mesquita and Boiger 

2014; Boiger and Mesquita 2015; Cunningham et al. 2015; Lindquist 2013), also referred to 

as a research programme, all sharing a common constructionist foundation (Russell 2015). 

This common underlying principle echoes Jamesian conviction that emotions are 

constructed, not engendered, out of more basic psychological primitives that are not 

emotion-specific but “domain-general ingredients from which experiences emerge more 

generally” (Barrett et al. 2015: 84). A notable premise of the psychological construction 

accounts is, further, that emotions are highly heterogeneous phenomena characterized by 

considerable variation that, following Barrett et al. (2015: 85), “is the key to survival”. 

By analogy to basic emotion and appraisal theories, there is a certain degree of 

variation in the formulations of and emphasis on some constructionist assumptions in 

different constructionist models. As already noted, constructionist accounts unite in the 

investigation of basic, psychological primitives underlying subjective experience, but differ 

in their delineation and formulation. For example, Russell’s (2003, 2009, 2012, 2015) 

psychological construction perspective proposes core affect (see section 1.3.2) as the most 



 26 

basic and fundamental property of a human mind, a psychological primitive that constitutes 

a vital ingredient of subjective emotion experience. Barrett (2012, 2014; Barrett et al. 

2015), in her Conceptual Act Theory, focuses more on how the interplay between basic, 

core systems (like core affect) gets constructed and conceptualized by the perceiver in what 

she refers to as situated conceptualization. In a yet different psychological construction 

perspective, Mesquita and Boiger (2014; 2015) argue that “emotions emerge from social 

interactions and relationships, which they in turn constitute, shape, and change (…). 

[S]ocial interaction and emotions form one system of which parts cannot be 

separated”(2014: 298). In this model, interpersonal, social, and cultural contexts become 

the core systems from which emotions are constructed. 

1.2.3.1. Critical evaluation 

It seems that the contemporary psychological construction models of emotion have so far 

managed to evade strong criticism, possibly because psychological construction accounts – 

in their current form – are relatively recent and some of them are still in the making. One 

valid objection to psychological constructionists’ denial of the existence of discrete neural 

circuits of basic emotions, however, was raised by a neuroscientist Joseph LeDoux (2012b). 

Specifically, LeDoux (2012a, 2012b, 2014, 2015) argues that despite rapid development in 

human neuroscience, the available neuroimaging techniques do not allow for detailed 

enough a picture of the neural correlates of basic emotions in humans; hence, due to such 

technological limitations, one cannot disprove the hypothesis proclaiming distinct, innate 

neural circuits of ‘fear’, ‘anger’, and other basic emotions (LeDoux 2012b), as argued by 

psychological constructionists (e.g. Barrett 2006a; Lindquist et al. 2012; Barrett and 

Satpute 2013).  



 27 

1.2.4. Does the concept of emotion serve any useful purpose in scientific psychology?
2
 

In her article entitled “Emotion: an example of the need for reorientation in psychology”, 

Duffy (1934a: 186) argued that “lack of success in recognizing emotion could be due to 

faulty experimental techniques and to inadequate guides for introspection, but it could also 

very well be due to the fact that the object of our search is, in the form in which we seek it, 

non-existent”. It has been 81 years separating Duffy’s publication from present 

investigations of emotion, a period marked by remarkable progress in research 

methodology. The improvement of behavioural and development of neuroimaging 

techniques have made it possible to indirectly measure neural correlates of processes 

involved in emotional experience. Despite such advances, however, the scientific 

understanding of emotion remains to be poorly understood and its definition constantly 

debated (LeDoux 2012b); this could be also inferred from the aforementioned discussion of 

different approaches to emotion. Hence, echoing the ideas of James (1884, 1890) and Duffy 

(1934a, 1934b), most recent approaches to emotion have focused on the investigation of 

more global non-emotional processes that underlie subjective experiences of emotion 

(Schachter and Singer 1962; Russell 2003, 2009, 2015; Barrett 2014; Barrett et al. 2015; 

LeDoux 2012b, 2012a, 2015; Clore and Ortony 2013; Scherer 2009; Moors et al. 2013; 

Panksepp and Watt 2011; Walla and Panksepp 2013), abandoning the common sense term 

‘emotion’ as an object of scientific investigation. In line with LeDoux (2012),  

[t]he challenge for emotion researchers is to understand the relation of the phenomena to the 

field of emotion without redefining them as fundamentally emotional phenomena, and thus 

infusing the phenomena with confusing implications. (…) Stepping back from the 

overarching concept of emotion and focusing instead on key phenomena that make emotion 

an interesting topic may be the best way out of the conceptual stalemate that results from 

endless debates about what emotion is. (LeDoux 2012b: 653–654) 

This does not mean that the term ‘emotion’ should be discarded altogether. It is indeed a 

fundamental ingredient of human interactions when used as a folk term to conceptualize 

and communicate the myriad of otherwise elusive bodily sensations. In the context of 

                                                 
2
A quotation from Duffy (1934a: 184). 



 28 

scientific investigation, however, it seems that the psychological construction analysis of 

component parts of emotional experience constitutes a more reliable and transparent 

scientific paradigm that might in the long run provide a comprehensive understanding of 

this phenomenon. Focus on a more general process (e.g. valence, arousal, body feedback) 

might therefore allow for more testable hypotheses, as well as more accurate and 

generalizable interpretations of findings in future research on such processes that contribute 

to the construal of subjective emotion experiences. 

1.3. Theoretical framework 

1.3.1. Psychological construction: a framework for the present investigation 

The psychological construction analysis of emotional phenomena constitutes a theoretical 

framework that, at present, most accurately accounts for the way in which emotional 

processes will be investigated in this dissertation. Specifically, consistent with 

psychological construction model, the present investigation does not focus on the analysis 

of the common sense, folk terms of emotions, but rather on more general affective 

processes and their realization and manifestation in language. One such process, valence, 

will be given particular importance, especially in the analysis of the affect-language 

interface. Valence is a term coined by a German-American psychologist, Kurt Lewin 

(1935), and refers to the intrinsic pleasant (positive/attractive) or unpleasant 

(negative/unattractive) quality of a stimulus, event or situation that arises in an interaction 

between an individual and their environment. From the perspective of a psychological 

construction model, valence, along with arousal (the level of intensity of physiological 

response to a stimulus, event or situation), constitute a fundamental ingredient of human 

emotion experience and the most basic affective substrate – core affect (Russell 1980; 

Russell and Barrett 1999; Russell 2003, 2009, 2012, 2015).  



 29 

In what follows, I will provide a theoretical account of the concept of core affect, 

which has emerged relatively recently, but at present might constitute a reliable and 

accurate theoretical framework for the analysis of the whole array of human emotional 

experiences. Following the discussion of core affect, I will proceed to the overview of the 

ubiquitous process accompanying human life – evaluation –, where more general affective 

processes play a key role. I will then review what is referred to as the affective primacy 

hypothesis, according to which affect dominates social interaction and thus is given priority 

in the course of processing. In this context, I will also try to tentatively propose the view 

that both evaluation and the processes underlying affect processing might be contingent 

upon people’s current state of core affect. The final part of this section devoted to 

theoretical and empirical analyses of affective phenomena will be devoted to a review of the 

neural correlates of affective processes from the perspective of psychological construction 

model. 

1.3.2. Core affect 

The primary motivation behind the attempt to formulate a new concept that might account 

for the wealth and variety of human emotional experiences was quite straightforward: first, 

the prototypical emotional episodes or so called full-blown emotions (e.g., anger, fear, 

sadness; see Russell and Barrett 1999) turned out to be too specific categories to do so 

effectively; second, they are constrained to the English language that in itself turned out to 

be very limiting when attempting to analyse cross-cultural emotion experiences (see 

Russell 1991, 1994). Thus, consistent with the psychological construction model there was 

a need for the re-analysis of the concept of emotion by focusing on more basic, primitive 

ingredients of emotional phenomena that might provide a more comprehensive 

understanding of human emotion experiences.  

The concept of core affect was born out of the analyses of people’s subjective 

reports about their feelings and moods in a given point in time (Russell 1979; Russell et al. 

1989; 2015). Such analyses demonstrated that how individuals felt might be reliably 



 30 

represented as a mixture of two general, independent, bipolar dimensions: valence 

(pleasant/unpleasant) and arousal (active/drowsy); this relation may be represented in a 

two-dimensional space, as demonstrated in Figure 1. According to Russell and Barrett 

(1999) a representation of an individual’s state of core affect requires that the two 

dimensions of valence and arousal be taken into account; experience-wise, however, core 

affect is a single feeling (Russell 2012). 

 

Figure 1. A representation of core affect in a two-dimensional circumplex model (after Russell 2012: 86) 

Core affect thus represents “the most elementary consciously accessible affective 

feelings” (Russell and Barrett 1999: 806). More precisely, it has been construed as “a pre-

conceptual primitive process, a neurophysiological state, accessible to consciousness as a 

simple non-reflective feeling: feeling good or bad [the dimension of valence], feeling 

lethargic or energized [the dimension of arousal]” (Russell 2009: 1264). In short, the state 

of core affect individuals find themselves in would be reflected in an answer to a question: 

“How do you feel?” (Russell 2015: 198). Notably, unlike full-blown emotions, core affect 

does not require an object; in other words, it can be “free-floating”, about nothing in 

particular (e.g. I feel good!), in which it resembles moods (Russell and Barrett 1999: 806; 

Russell 2003, 2015). If it does get attributed to an object (e.g. Your presence makes me feel 

good!), however, it then constitutes one of the sub-events of prototypical emotional 



 31 

episodes. Hence, core affect is thought to be the basic affective substrate that constitutes an 

essential ingredient of full-blown emotions. 

Another characteristic feature of core affect is that it “ebbs and flows” as a function 

of time (Russell and Barrett 1999: 806). Such fluctuations in core affect vary in degree and 

may have important implications for other cognitive processes. For instance, a person being 

in a state of negative core affect may be more likely to perceive and appraise their 

environment in a more negative light. Other cognitive processes, such as decision making, 

have been also argued to be contingent upon the current state of core affect (see Russell 

2015). In fact, it has been demonstrated that core affect modulates participants’ behavior 

even if it is not consciously experienced (Berridge and Winkielman 2003; Winkielman and 

Berridge 2004; Winkielman et al. 2005). Hence, according to Barrett (2006b: 50) core 

affect is a “neurophysiologic barometer of the individual's relationship to an environment at 

a given point in time [emphasis mine, RJ]”. In other words, it is thought to be the governor 

of our perceptions and behaviours. As such it is considered a psychological primitive; a 

concept that is irreducible at the psychological level (Russell 2012). This has two 

implications. First, as the core characteristic of a psychological primitive is that it is not 

specific to any particular domain, so is the case with core affect and the domain of emotion. 

Specifically, it has been shown that core affect may also play a significant role in other 

non-emotional processes such as vision (see Barrett 2011). Second, it should be brought to 

the fore, that, although irreducible as a psychological phenomenon, core affect can be 

broken down into more elementary processes on the neural level of analysis. Although the 

investigation of core affect in the brain is a relatively recent development, it already 

provides initial evidence that individuals’ self-reports of their core affective states are 

strongly correlated with the neural activation in specific brain areas; activation that is not 

common to various instances of prototypical emotional episode (see Wilson-Mendenhall et 

al. 2013). 



 32 

1.3.3. Affective evaluations 

The neurophysiological state of core affect might play a significant role in human social 

interactions (Zajonc 1980; Kopytko 2002)
3
. While the research on the relation of core 

affect and other psychological domains is still in its infancy, there is introspective and 

objective evidence to tentatively suggest that individuals are constantly and often 

unconsciously (Berridge and Winkielman 2003, 2003) influenced by their affective states 

that guide their actions and behaviours. One important domain guided by core affect is the 

domain of evaluation, an indispensable part of people’s everyday experience (Barrett 

2006b: 38). Evaluation is thought to arise when the free-floating, neurophysiological state 

of core affect gets attributed to an object in the environment (e.g. another individual, an 

event, a situation). In other words, the process of evaluation may be construed as an 

individual’s feeling about a particular stimulus (Russell and Barrett 1999). People evaluate 

their environment continuously and effortlessly, consciously and unconsciously. Hence, 

according to Winkielman et al. (2003: 189) each human being has become ‘an evaluating 

human’- homo evaluaticus. Governed by core affective states, evaluation is essentially 

about an organism’s binary decision what is pleasant or unpleasant (valence), valuable or 

worthless (appraisal), and is automatically linked to approach-withdrawal behaviours (Chen 

and Bargh 1999: 220). Evaluation may be also manifested in what is often referred to as a 

‘gut feeling’ (Gigerenzer 2008). It often happens that we spontaneously decide to choose a 

restaurant, having already discarded many alternative options, but fail to provide a good 

reason for it. We somehow ‘like’ or ‘prefer’ one over the others. Or we simply say that we 

‘feel it in our gut’. This ‘gut feeling’ is probably what Russell (2003, 2009, 2012, 2015) 

might refer to as core affect. Overall, evaluations are thought to be primarily based on the 

interaction between a current state of core affect, a stimulus in the environment to which 

core affect is directed, and, crucially, the wealth of an individual’s experience.  

Extensive scientific investigation of evaluative processing in the fields of social and 

psychological sciences has demonstrated that such processing is fairly automatic (see Bargh 

                                                 
3
However, both Zajonc (1980) and Kopytko (2002) refer to more general affective phenomena rather than 

core affect (see Russell and Barrett 1999). 



 33 

1997; Bargh et al. 1992, 1996; Chen and Bargh 1999; Bargh et al. 2012; Williams and 

Bargh 2008). Interestingly, a number of studies have demonstrated a uni- as well as bi-

directional influence of the experience of physical warmth or coldness on how individuals 

perceive and interpret interpersonal relationships and use of language (see Williams and 

Bargh 2008; IJzerman and Semin 2009, 2010; Zhong and Leonardelli 2008). Specifically, 

the experience of increased physical warmth induced by holding a warm beverage (e.g. 

IJzerman and Semin 2009; Williams and Bargh 2008), staying in a warm room (IJzerman 

and Semin 2009), or taking a warm bath (Bargh and Shalev 2012) resulted in participants’ 

exhibiting prosocial behaviour (Williams and Bargh 2008), using more concrete language 

in their narratives – which is reflective of greater social proximity (see IJzerman and Semin 

2009) –, and even compensating for the feelings of social distance, exclusion and loneliness 

(Bargh and Shalev 2012: 12). This pattern of findings was reversed in the coldness 

condition. Of note, all these processes have been reported to take place outside of 

participants’ awareness. 

Although not interpreted in the core affect paradigm, this body of research could 

provide evidence in support of the hypothesis that fluctuations in the core affective state 

(here, reflected by alterations in the physiological measure of body temperature) not only 

influence subsequent evaluations or impressions of the environment but may also be 

consciously or unconsciously regulated
4
 to achieve a desired state or feeling (Russell 2003, 

2012; Bargh and Shalev 2012). Furthermore, the study by IJzerman and Semin (2009) 

demonstrated that alterations in the experience of physical warmth have a bearing on the 

use of language. It would be interesting to investigate this phenomenon further or to see if 

the reverse relationship would also be true: Could language, just like physical warmth, be 

considered a factor influencing a state of core affect? A common sense answer might be 

positive. Language is a communicative tool that, among others, is used to express how 

people feel (and thus in a way reflecting their current core affective state), but it may also 

be a reason why they feel that way. For example, the core affective barometer (Barrett 

2006a) might undergo dynamic fluctuations when reading poetry, hearing a compliment, a 

                                                 
4
In fact, Russell (2003, 2012, 2015) discusses individuals’ ability to modulate their current state of core affect 

through, what he refers to as, affect regulation. Russell argues that “[o]ne can seek to alter or maintain Core 

Affect directly – affect regulation – from the morning coffee to the evening brandy” (Russell 2015: 196). 



 34 

joke, a reprimand or an insult. Hence, the relationship between language and core affect 

might constitute an interesting area of future empirical investigation. 

In sum, research reviewed in this subsection aimed to highlight the omnipresence of 

evaluations in people’s everyday life (for a detailed discussion see Bargh et al. 2012; Bargh 

1997; Bargh and Chartrand 1999). Although this extensive body of research is not typically 

interpreted in the core affective paradigm, there might be a relationship between evaluative 

processes and the individual’s core affective state. At this point, however, this issue awaits 

and requires empirical verification. 

A more established and common view is that evaluative processes are governed by 

general affective processes (Winkielman et al. 1997). Affective information encoded in 

previous experience and stored in memory as well as – possibly – an individual’s current 

core affective state together constitute the foundation for evaluation
5
. In such a view, 

cognitive processing is engaged in evaluative processing to a very limited - if any – extent 

(Winkielman et al. 1997, 2003, 2005; Winkielman and Berridge 2004; Zajonc 2000). This 

limited role of cognitive resources relative to affective resources has been the main issue in 

the debate on affective primacy hypothesis put forward by Robert Zajonc in his 1980’s 

paper that created a real turn in affective sciences (Winkielman 2010: 357). 

1.3.4. Affective primacy hypothesis 

The main premise of the affective primacy hypothesis is that affective processing is 

primary and indifferent to cognitive processing (Zajonc 1980: 153). While feeling and 

thinking should not be seen as inseparable from each other, according to Zajonc (1980: 

154) affect is always a faithful companion of thought, but the reverse is not always the 

case.  

This view has been corroborated by evidence from multiple studies on the mere-

exposure effect (also mere-repeated-exposure effect) and subliminal affective priming (e.g. 

                                                 
5
This view has been commonly referred to as feelings-as-information model (Murphy and Zajonc 1993; 

Winkielman et al. 1997). 



 35 

Kunst-Wilson and Zajonc 1980; Monahan et al. 2000; Murphy et al. 1995; Zajonc 1968; 

Murphy and Zajonc 1993; Winkielman et al. 1997, 2005; Berridge and Winkielman 2003; 

Barrett and Bar 2009). In a mere-repeated-exposure paradigm, researchers measure a 

participant’s unfolding preference for a stimulus following repeated exposures 

(supraliminal or subliminal) to that stimulus. In short, with increased exposure to a stimulus 

comes increased liking and familiarity of that stimulus, independent of recognition. In 

mere-exposure experiments, participants are not engaged in any additional tasks, and very 

often not even aware of the presented stimuli (Zajonc 2001). Kunst-Wilson and Zajonc 

(1980) reported that affective discrimination can occur with minimal or no access to 

cognition. Their experiment consisted of 2 parts: first, participants were exposed to 

octagons flashed for a brief period of time; second, they were asked to recognize and 

evaluate the octagons. There were 2 sets of octagons: set A and set B, consisting of 10 

octagons each. In the first part of the experiment, a participant was exposed to one of the 

sets only. During the second part they viewed both sets of octagons in a comparative 

manner and were asked (1) to indicate the octagon they liked more and (2) to indicate the 

octagon they remembered. Despite very low (close to chance) recognition, affective 

recognition was preserved, providing evidence in favour of mere-exposure effect with no 

cognitive mediation (Kunst-Wilson and Zajonc 1980: 558). In a subliminal affective 

priming paradigm, researchers investigate whether and how subliminal presentation of an 

affective stimulus (i.e. an affective prime) has a bearing on the processing of the 

subsequently presented target stimulus. In an affective priming study, Murphy and Zajonc 

(1993) conducted a series of 6 experiments to directly investigate the interaction between 

affective and cognitive processing. All experiments provided strong evidence in favour of 

the affective primacy hypothesis, whereby affective valence of primes (facial stimuli) 

modulated the preference for target stimuli (Chinese ideographs) in the subliminal priming 

condition (stimulus exposure = 4 ms), while non-affective features of primes (e.g. shape, 

symmetry) influenced participants’ likeability ratings only in the supraliminal priming 

condition (stimulus exposure = 1000 ms; for details, see Murphy and Zajonc 1993). These 

results may be interpreted in line with the hypothesis that affect is primary and influences 

subsequent cognitions. The study by Murphy and Zajonc (1993) also demonstrated that 



 36 

human beings have the ability to make valence discrimination (good vs. bad) outside of 

consciousness, which is possibly a trace of evolution.  

The affective primacy hypothesis was criticized by Richard Lazarus – the proponent 

of cognitive primacy hypothesis (see Lazarus 1984, 2006). Contrary to Zajonc (1984), 

Lazarus (1984) claimed that cognition was a prerequisite for affective reaction to occur, to 

quote: “[c]ognitive activity is a necessary precondition of emotion because to experience an 

emotion, people must comprehend (…) that their well-being is implicated in a transaction, 

for better or worse” (1984: 124). What should be noted, however, is that Zajonc (1984) did 

not proclaim, what Lazarus interpreted as (1984), “the independence of cognition and 

emotion and the primacy of emotion [emphasis in the original, RJ]” (1984: 124). Instead, 

Zajonc’s research focused on the primacy of affective processing, organism’s primitive 

mechanisms that enable instant evaluations of a stimulus (positive vs. negative) in the 

environment to guide its behaviour (approach vs. withdraw). Even if at that point in time 

Zajonc did not really differentiate between affect and emotion, his focus was mainly on 

investigating how subliminal perception of repeated exposure to affective stimuli could 

influence instant preferences for that stimulus. Such basic evaluations or preferences may 

happen without the company of cognition as demonstrated by the substantial evidence 

supportive of this claim (see Zajonc 2000), which was also reviewed in the previous section 

on evaluative processes. Lazarus’s (1984) rebuttal might not seem to be valid as it looks as 

if he built his counter-arguments on the foundation of emotion experience, not affect 

perception. Therefore, the tension between the two scholars might arise from a fundamental 

difference in their theoretical and empirical approaches to affect and emotion, an issue of 

terminology that – as already argued in the introductory sections – permeated theoretical 

and empirical research on affect and emotion resulting in misinterpretations, ambiguity and 

hindering further progress in the field. Hence, affective primacy hypothesis and the 

cognitive primacy hypothesis should not be necessarily seen as opposing theories; to the 

contrary, they could complement each other, when understood as different stages of affect 

perception and experience. 

In sum, the debate between Zajonc and Lazarus provoked many questions that 

translated into further quests for investigating the causal mechanisms underlying affective 



 37 

reactions. Some researchers raised questions related to the potential existence of 

unconscious affect that would impact behaviour, but whose elicitation would not require the 

medium of consciously experienced feelings. This line of research is the topic of the next 

section. 

1.3.5. Unconscious affect 

Emotions cannot be unconscious because they must be felt, and feelings are by definition 

conscious. (Clore et al. 1994: 290) 

It is surely of the essence of an emotion that we should feel it, i.e. that it should enter 

consciousness. (Freud 1950: 109–110) 

Research by Robert Zajonc demonstrated that the underlying, causal mechanisms of affect 

and affect assignment may function on the unconscious level. At the same time, however, 

Zajonc argued that the affective state, when elicited, was always assumed to be experienced 

consciously, as a feeling or emotion (Winkielman 2010: 358; Berridge and Winkielman 

2003: 185; see Zajonc 2000); in that, he would possibly agree with Freud (1950) and Clore 

(1994). In a similar vein, William James (1913) argued that a feeling or emotion – even if 

elicited on the unconscious level – is eventually experienced and enters human 

consciousness. Overall, the debate on whether affect or emotion might be processed 

unconsciously has had a strong presence in this field of research, particularly since the 

Zajonc-Lazarus (Zajonc 1984; Lazarus 1984) confrontation (see Prinz 2004). 

Based on already reported findings implicating the independence of the affect 

system from the cognitive system, some researchers set out to investigate if subliminal 

affective priming may have an impact on subsequent cognitive processing without the 

elicitation of conscious feelings (e.g. Winkielman et al. 1997, 2005; Berridge and 

Winkielman 2003). For example, Winkielman, Zajonc, and Schwarz (1997) investigated 

participants’ preference for Chinese ideographs in a covert affective priming paradigm with 

happy or angry faces. Participants in the study were purposely informed that the ideographs 

would be preceded by a negative or a positive face (experiment 1) or that the music playing 



 38 

in the background might elicit a positive or negative mood (experiment 2; Winkielman et 

al. 1997: 442; Winkielman and Berridge 2004: 121). Hence, they were asked to pay 

attention to their feelings, but not to rely on them while expressing preferences for 

ideographs. The experiments revealed that the affective priming effect was still present 

irrespective of the attributional manipulations. Furthermore, post-experiment interviews 

with participants revealed that they did not experience changes in affective states to primes 

throughout the experiment, but did so in response to music (Winkielman et al. 1997: 456). 

This led the authors to hypothesize that basic preferences may be influenced by 

unconscious affect elicited by subliminally presented affective faces (Winkielman et al. 

1997: 462). As suggested by Winkielman and Berridge (2004: 121), however, the reason 

for participants’ not having reported any affective changes in response to primes in the 

study might be attributed to simply not remembering the feelings during a post-study 

interview. To address this issue in further experiments Winkielman et al. (2005) collected 

participants’ self-reported ratings of mood and stimuli directly after the presentation of 

subliminal affective primes. The authors demonstrated that subliminal presentation of 

affective primes had an influence on the amount of beverage consumed and the beverage 

ratings. These changes in behaviour did not translate into changes of self-rated mood, 

which was interpreted as evidence in favour of the existence of unconscious affect 

(Winkielman et al. 2005: 132). These results show that basic affective processing 

influences not only basic behaviour such as whether to approach or withdraw, but also 

more complex behaviours such as consuming a beverage. Furthermore, this study extended 

previous research showing that “preferences need no inference” (Zajonc 2000) and 

demonstrated the affect-motivation interaction (Winkielman et al. 2005: 132). In sum, the 

study by Winkielman et al. (2005) was possibly the first to provide preliminary evidence 

that unconscious affect may have a direct impact on human behaviour (Winkielman et al. 

2005: 133). 

The discussion on unconscious affect or emotion has also strongly resonated in the 

research on animals (Panksepp 1998; Walla and Panksepp 2013; LeDoux 2012b). 

Specifically, it has been argued that since rodents and non-human primates exhibit basic, 

instinctive affective or emotional responses, such as freezing or fleeing in response to 



 39 

threat, by analogy, this evolutionary trace would also be preserved in humans in similar 

contexts (LeDoux 2012a, 2015). LeDoux (2012b, 2014, 2015), however, cautions against 

applying subjective human experiences to animals calling such responses emotional or 

affective, and instead describes them as basic, innate survival functions or responses guided 

by survival circuits that “did not evolve to make emotions but to give organisms 

behavioural tools to stay alive” (LeDoux 2014: 319). As such, basic survival responses 

(e.g. seeing a bear in the woods
6
) may be relevant to emotion, but they should not be 

considered direct causes of emotions. Hence, the unconscious responses to threat or danger 

in the event of seeing a bear in the woods or a snake in the grass should not be viewed as 

unconscious instances of fear, as argued by other animal researchers (see Panksepp 1998; 

Walla and Panksepp 2013; Panksepp and Watt 2011), but rather as more elementary, non-

emotional survival responses. These arguments show that the debate on whether 

unconscious affect or emotion exist might again to a significant extent depend on how 

researchers define the phenomena in question.  

1.3.6. The neuroscience of affect 

there is no ‘affective brain’, ‘social brain’, or ‘cognitive brain’. Each human has one brain 

whose functional properties can be understood differently for different time scales and levels 

of organization. (Barrett and Satpute 2013: 368) 

Until recently, the leading idea in neuroscience was that certain cognitive faculties like 

language or emotion were meant to be generated and localized in specific parts of the brain 

that, by analogy, would specialize in the processing of specific, linguistic or emotional, 

information, respectively. This idea was postulated by the faculty psychology framework 

(e.g. basic emotion theory) that views common sense experiences in terms of basic faculties 

of the mind (see Lindquist and Barrett 2012). This framework has been prevalent in 

affective neuroscience for the recent 30 years, whereby researchers would set out to 

discover the brain loci of basic emotion categories (e.g. where fear is located in the brain?), 

                                                 
6
 The famous example used by William James (1884, 1890). 



 40 

looking for evidence in favour of the innate and distinct neural correlates of each emotion 

category. And so, early findings from neuroscience of emotion hailed amygdala as the host 

and generator of fear (Sprengelmeyer et al. 1999; see Vytal and Hamann 2010), anterior 

insula as the host and generator of disgust (Wicker et al. 2003; see Vytal and Hamann 

2010), subgenual anterior cingulate cortex (sgACC) as the host and generator of sadness 

(see Murphy et al. 2003), and rostral ACC and right superior temporal gyrus (STG) as the 

hosts and generators of happiness (see Vytal and Hamann 2010).  

Most recent evidence from neuroscience, however, does not confirm the previous 

predictions of the faculty psychology approach to emotion. Contemporary research 

demonstrates that brain regions previously identified with generation of particular emotions 

are also engaged in a wide array of different mental processes (see Barrett and Satpute 

2013; Lindquist and Barrett 2012; Lindquist 2013). To give but one example, the amygdala 

–commonly perceived as the centre of emotion processing in the brain– is also activated in 

response to novel stimuli that are not in any way affectively loaded (Weierich et al. 2010; 

Moriguchi et al. 2011), with the activation being comparable to that elicited by stimuli 

valence or arousal (Weierich et al. 2010: 10). Furthermore, more often than not brain 

regions associated with the processing of a specific emotion category (e.g. amygdala and 

fear processing) failed to elicit consistent neural activation in response to that emotion, or, 

by contrast, became activated in response to more than one emotional category (for 

evidence see Lindquist et al. 2012). In line with the psychology construction model, these 

findings contribute to the hypothesis that the brain does not respect categories constructed 

by the human mind, such as emotion, cognition, or perception (Lindquist et al. 2012: 138). 

Instead, what seems to be – at this point in time – a more evidence-driven explanation is 

that the human brain responds primarily to more primitive, basic processes (i.e. 

psychological primitives) that elicit a mixture of activations and deactivations of complex 

and large-scale distributed neural networks (and connections between those networks) with 

no single locus that would be selectively activated to a specific kind of stimulus (Power et 

al. 2011; Tomasi and Volkow 2011; Crossley et al. 2013; Yeo et al. 2011). LeDoux (2014) 

has recently commented on this issue by means of a ‘soup’ metaphor: 



 41 

With regards to conscious emotions (feelings), my view (…) can be summarized by way of 

analogy to the way the character of a soup arises from its ingredients. None of the ingredients 

are soup ingredients. They are things that exist in nature and that can be used in soups of 

various kinds, and in other kinds of dishes as well. But the particular combination of 

ingredients gives the soup its character. Similarly, nonemotional ingredients (…) come 

together to give rise to an emotion, a feeling. (LeDoux 2014: 319) 

The psychological construction view of the affect-brain interaction also echoes the 

arguments of William James (1913) who questioned the idea of the existence of special 

neural centres dedicated to emotion. Such an approach seems to be not only more intuitive, 

but also more explanatory of the incredible economy and holism of the brain’s organization 

and function (Bullmore and Sporns 2012). In what follows, I review the most recent 

findings concentrating on the neural correlates of core affect and more general affective 

processes from the more holistic perspective offered by the psychological constructionism 

model. Specifically, I will focus on the most recent neuroscience evidence outlining the 

neural networks engaged in the affect processing as well as the interplay between affective 

and social processes in the brain. 

In a recent and influential resting-state functional connectivity MRI (rs-fcMRI)
7
 

study among 1,000 participants, Yeo et al. (2011: 1135) demonstrated that the human 

cerebral cortex consists of a collection of 7 large-scale distributed association networks 

characterized by highly interconnected regions. These networks include the visual, 

somatomotor, dorsal attention, ventral attention (also referred to as the salience network), 

limbic, frontoparietal, and default networks (for illustration and description of brain regions 

included in each of the networks, see Figure 2). Some or all of those networks have been 

already reported in previous investigations of neural networks in the human brain (e.g. 

Beckmann et al. 2005; Damoiseaux et al. 2006; De Luca et al. 2006; Fox et al. 2006; 

Vincent et al. 2008; Buckner et al. 2008). Importantly, four of the aforementioned networks 

have been associated with affective processing, namely the salience network, the 

frontoparietal network, the default network, and, to a lesser degree, the limbic network (e.g. 

Oosterwijk et al. 2012; Lindquist 2013; Barrett and Satpute 2013). For example, in an fMRI 

                                                 
7
Resting-state functional connectivity MRI (rs-fcMRI) enables to investigate spontaneous fluctuations in brain 

activity during resting state, i.e. when participants are not asked to perform any explicit task (Fox and 

Greicius 2010; Yeo et al. 2011).  



 42 

study, Oosterwijk et al. (2012) measured participants’ brain activity within large-scale 

distributed networks in response to scenarios that were used to evoke emotions, body 

sensations or thoughts (for details see Oosterwijk et al. 2012: 4). The study showed that all 

mental states elicited activation in the salience, frontoparietal, default, and, less so, limbic 

networks (Oosterwijk et al. 2012: 14–15).  

 

Figure 2. The visualization of the 7 brain networks (after Yeo et al. 2011: 1137), with an added description of 

the regions included in each of the network (see Yeo et al. 2011; Oosterwijk et al. 2012) 

Similar findings have been also reported in meta-analyses of neuroimaging studies 

on affect and emotion (Lindquist et al. 2012; Kober et al. 2008; Lindquist and Barrett 2012; 

Barrett 2012). A recent meta-analysis of 397 fMRI and positron emission topography 

(PET) studies found that positive and negative affect is processed in shared neural 

networks, with no distinct region for positivity or negativity in the brain (Lindquist et al. 

2015). These findings show that affect processing elicits activation within widely 

distributed neural networks and, notably, that similar neural networks are engaged in the 

processing of other, not necessarily affect-related, mental states. Indeed, a recent meta-



 43 

analysis using the activation likelihood estimation (ALE)
8
 reported that both socio-

cognitive and emotional tasks employed in the studies selected for analysis elicited 

activation within a common neural network consisting of dorso-medial prefrontal cortex 

(DMPFC), middle and anterior superior temporal gyrus (STG), the precuneus, and the 

ventrial striatum (Schilbach et al. 2012: 6). This finding might provide evidence for a 

significant role of affect in social cognition. In a similar vein, numerous studies 

demonstrated that the cognitive and affective neural networks largely overlap (Shackman et 

al. 2011; Raz et al. 2012, 2014; Duncan and Barrett 2007). For example, the dorsolateral 

prefrontal cortex (dlPFC) that until recently was perceived the ‘cognitive’ centre of the 

brain has been also demonstrated to be engaged in top-down control of emotion and 

motivation (see Okon-Singer et al. 2015: 15). Finally, control processes have been thought 

to play an important role in the context of threat, whereby there is a necessity to monitor 

risk, and the unfolding of a potentially dangerous action. This is also reflected in the shared 

neural networks of negative affect and cognitive control in the midcingulate cortex (e.g. 

Shackman et al. 2011; Lin et al. 2014; see Okon-Singer et al. 2015: 17). 

In sum, contemporary research in the field of neuroscience has adopted a more 

holistic, network-based approach to the investigation of neural correlates of emotion. In 

doing so, the focus in the research shifted from analysing specific and distinct loci in the 

brain where emotion categories were believed to be generated and processed (aka the 

faculty psychology framework) to investigating activation of large-scale distributed neural 

networks in response to psychological primitives, or “lowest common denominators” (aka 

the psychology constructionism framework; Lindquist and Barrett 2012: 5) that are thought 

to lie at the core of the mental states we experience on the surface. 

Evidence from neuroscience demonstrating shared neural correlates of affect and 

social processing may provide yet another vital information about the nature of affect, i.e. 

that affect is by nature ingrained in social interaction (see Schilbach et al. 2012). Robert 

Zajonc (1980) has proposed that “affect dominates social interaction” (1980: 153; for a 

similar view see Kopytko 2002); furthermore, affect is generally thought to be shaped by 

                                                 
8
The ALE approach enables to statistically analyze activations and deactivations of selected brain regions 

across published neuroimaging reports (Schilbach et al. 2012: 1). 



 44 

personal experience and social interactions, and therefore is not seen as a stimulus property 

(Barrett 2006b: 50).  

One of the primary mediums of communicating affect is language; both its non-

verbal and verbal aspects. In the following sections I will provide an overview of the 

interaction between affect and language from two different perspectives: (1) a 

psycholinguistic perspective, which has been very productive in explaining the psycho- and 

neuro-mechanics of how people process affective language, but did not consider context an 

important variable in such investigations and thus decreased their ecological validity 

(section 1.4.1), and (2) a pragmatic perspective, which so far has produced only a few 

studies on the affect-language interaction, but in doing so it has provided first insights into 

how people might process affective language in everyday communicative interactions 

(section 1.4.2). 

1.4. Affective processing of verbal stimuli 

Affect permeates the entire linguistic system. Almost any aspect of the linguistic system that 

is variable is a candidate for expressing affect. In other words, language has a heart as well as 

a mind of its own. (Ochs and Schieffelin 1989: 22) 

Common sense dictates that affect and language are closely connected, if not inseparable 

phenomena (Moeschler 2009). Affect is smuggled in everyday linguistic interactions (both 

speech- and text-based), and its presence particularly reveals itself in the melody of 

language (prosody), poetry or in the use of taboo words and swearwords. As argued by 

Besnier (1990: 421), “affect floods linguistic form on many different levels of structure in 

many different ways”. Despite the proclaimed pervasiveness of affect in language, 

however, this topic has been understudied in the field of mainstream linguistics
9
 (see 

Besnier 1990; Reilly and Seibert 2003; Langlotz and Locher 2013; Jay and Janschewitz 

2007) that focused mainly on the denotative (rather than connotative) meaning of language 

                                                 
9
Here, I mainly refer to structural and generativist approaches to linguistics. Some functionalist approaches to 

linguistics, by contrast, have made initial attempts to include the emotive or expressive function of language 

in their analyses (Bühler 1934; Jakobson 1960). 



 45 

and its purely cognitive, non-emotional nature. Indeed, the relationship between affect and 

language has been addressed more often in the fields of anthropology (e.g. Malinowski 

1923; Besnier 1990) and, even more so, psychology (to be reviewed below).  

In the coming subsections, I will start by reviewing the body of research from the 

field