A place for prosody in a unified model of cognition and emotion

Full text

A Place for Prosody in a Unified Model of Cognition and Emotion
Branka Zei Pollermann
Liaison Psychiatry
University Hospitals of Geneva, Switzerland
branka.zei@hcuge.ch
Abstract
The relationship between cognition and emotion has been one
of major stumbling blocks in defining the concept of emotion.
We argue that a unified model of cognition and emotion based
on interactionist epistemology would allow to unambiguously
define emotion and thus facilitate research in affective aspects
of speech communication. The main tenet of the model is that
cognition itself includes three dimensions of affective space:
valence, arousal and coping power. The semiological status of
prosody is discussed and two examples of our own research
are given.
1. Introduction
Over the past hundred years, the psychology of emotions has
mainly concentrated on various components of the organism's
emotional response such as: cognitive, physiological,
neurological, motor, expressive, behavioral, linguistic, and the
subjective feeling.
We are currently facing a remarkable diversity of
emotion theories, each focusing on a different component of
the organism's response system. Most theories largely overlap
in terms of hypothesized or empirically found components of
emotions. Yet, there is no consensus on the criteria for
distinguishing between emotional and non-emotional
phenomena inherent to human behaviour.
This situation is well illustrated in The Nature of Emotion
[11] where twenty-three eminent scholars answered twelve
fundamental questions regarding the nature of emotions. In
the concluding chapter, the editors deny the existence of a
gold standard for knowing that an emotion has occurred.
Nonetheless the contributors seem to agree that emotion can
be characterised as a phenomenon which involves organism
totally, including the shaping of personality. Now, the
question remains: how do we distinguish between emotional
and non-emotional responses? Scherer [32] believes that the
difference rests upon one distinctive feature: temporary
synchronisation of the functioning of all or most of the
organismic subsystems which occurs when the response is
emotional. However, we believe that most adaptive reactions
to relevant stimuli require a certain degree of synchronisation
of the work of organismic subsystems without necessarily
being recognised as emotional.
One of the stumbling blocks in defining emotion has been
the relationship between emotion and cognition. We believe
that a change of paradigm is required in that the question must
be addressed from an epistemological point of view. Unlike
some scholars [15] we believe that the relationship between
cognition and emotion is a fundamental epistemological issue
and not just a semantic controversy.
The model we are proposing is based on two
complementary epistemological approaches: Piaget's genetic
epistemology [22] and Prieto's semiological theory of
knowledge [25]. We shall thus first present an outline of a
unified model of cognition and emotion, which will allow us
to distinguish between emotional and non-emotional
reactions, next we shall address the question of the
semiological status of prosodic indicators of affective-
cognitive states in terms of: expression vs. communication of
affect, and we shall end with two examples from our own
research.
2. Cognition is in emotion and emotion is in
cognition
It is now generally accepted that for an emotion to occur, the
subject must cognitively evaluate the significance of the
occurring external or internal stimuli. Most often, this is done
unconsciously. It is also accepted that cognition plays a part
not only in the appraisal of the stimulus [13] [27] [30] but also
in the awareness of the overall subjective feeling [29]. On the
other hand, there has been evidence [7] that emotional
processes are required for certain types of decision-making to
occur. In addition, studies in cognitive neuroscience of
emotion [12] point out that “there is no evidence that neural
processes subserving cognition differ in fundamental ways
from those subserving emotion” [12 p.5].
It thus appears that the relationship of cognition to
emotion needs clarification. We believe that this can be
accomplished within a unified theory which integrates
emotion, cognition and behaviour into one model of the
organism's adaptive functioning.
2.1. General Theoretical Background of the Model
The proposed model applies to organisms described as open
systems with dynamic teleology [2] which have the following
characteristics:
1. A system is defined as a composition of elements in
interaction which is not random.
2. Behaviour is determined by the foresight of the goal.
3. The system is characterised by equifinality i.e. the same
state can be reached from different initial conditions and
in different ways.
4. The system has self regulatory features which aim at
maintaining a steady state.
5. Temporary changes in the environment provoke
reversible fluctuations of the organism's steady state.
6. The system is an intrinsically and autonomously active
system with its parts and processes being in mutual
interaction.
The main tenet of a unified model is that cognition is a process
whose mechanism necessarily includes the three basic
dimensions traditionally attributed to emotions: valence,
arousal and power. In other words, emotional dimensions are
inherent to cognition. The model is best understood when
Speech Prosody 2002
Aix-en-Provence, France
April 11-13, 2002
ISCAArchive
http://www.isca-speech.org/archive
10.21437/SpeechProsody.2002-3

placed in the context of interaction between the subject and
the environment.
Given that all interaction requires the use of knowledge of
the outer world and that of self, the model posits that both
kinds of knowledge form a semiotic structure which subserves
the subject's adaptive behaviour: the knowledge of the outer
world is constructed with regard to its significance for the
subject's interests and behaviours [25]. Ultimately, all
adaptive behaviour aims at: maintaining or achieving a
positive state (positively valenced state) and/or avoiding a
negative state (negatively valenced state). It is assumed that
the target states can concern any or all of the three dimensions
of Self: intra-subject (personal interests, including own body),
inter-subject (relational and social interests), and trans-
subject (interests related to the nature as a whole, or religious
pursuits).
The organism's cognitive and sensori-motor functioning,
can constitute positive target states on their own. These are
often related to creativity in scientific or aesthetic pursuits and
to self-realisation in sports performances. According to the
unified model, the inner mechanism of the cognitive process
involved in adaptive interaction - in which the subject
identifies the stimulus, and constructs relevant knowledge
about it - can be described as follows:
In the initial phase of stimulus perception and
identification of its characteristics, the subject constructs or
evokes the already existing knowledge about the stimulus and
its identity. This knowledge is attributed a value in terms of
its significance for the subject [24], including the intra, the
inter and the trans subject dimensions. This value can be
conveniently described as a value-tag*. We hypothesise that
the process of stimulus identification and its appraisal
includes attribution of three kinds of value-tags as three
dimensions of cognitive-affective space. Each value tag can
be conceived of as a scalar ranging from positive to negative
values.
∑ A first value-tag, termed valence tag, is attributed with
regard to the appraised beneficial or detrimental character
of the stimulus. This kind of value is considered as the
exchange value because it is related to what it can be
exchanged for, as far as the subject's wellbeing is
concerned. Caccioppo and colleagues [4] showed that the
function of valence tags is to regulate
approach - avoidance behaviour. Valence tags thus
represent the subjective meaning that the stimuli have for
each of the constituents of Self. Valence tagging is
considered to be inherent to the subject's interactive
behaviour
∑ b) A second value-tag, termed power tag, is related to the
subject's estimate of his/her coping and/or decisional
power. Prieto attaches a great deal of importance to this
point, which he links to the constitution of Self. A power
tag thus denotes the degree of the subject's power to
decide and to act upon the stimulus and/or its
consequences. It refers to the estimated relative power i.e.
the relation between available power and the power
needed to cope with the situation.
∑ c) A third type of value-tag, termed activation tag,
denotes the amount of energy expenditure involved in
motoric and autonomic-physiologic changes
automatically triggered or estimated as required to handle
the stimulus and/or its consequences. It is related to the
regulation of bodily, and computational energy
expenditure. Piaget calls this kind of value “valeur de
rendement” - the yield value which is also influenced by a
kind of cost-benefit calculus [23 p.595]. Piaget considers
bodily energy management to be an affective dimension
of behaviour.
* The term “tagging” has been borrowed from Ohman [19]
who uses it to denote the mainly unconscious process of
assigning emotional meaning.
In summary, each moment's implicit and explicit knowledge of
the environment, of self, and of one's own action, can be
placed into a three dimensional affective space of valence,
power, and arousal. There is reason to believe that valence
value is primary [8] in that it recruits and drives responses
including the appraisal of power, and arousal. In summary, our
unified model suggests that cognitive processes involved in
interaction with environment necessarily include dimensions
which are typically known as emotional, that is: valence,
activation, and coping /decisional power. The inseparability
of cognitive and affective aspects of adaptive action is clearly
expressed by Piaget: “There is no reason to separate
intelligence from affectivity, nor is it necessary to ask which of
the two precedes and conditions the other. It is a matter of two
inseparable aspects of mental development.” [24 p. 41].
Inspired by insightful works of Elisabeth Duffy in the
physiology of behaviour [10], and more recent work in
neurological aspects of emotions [5] we suggest the following
criteria for distinguishing between emotional and non-
emotional organismic responses:
If the organism's monitoring system reports critical values
(that is too low or too high) for valence, arousal or power, the
state becomes cognitively dominant. Such a moment's
dominant body-mind landscape is likely to penetrate
consciousness and be conceptualised as emotional. This is in
agreement with Ledoux's proposal regarding the neurological
triggers of consciousness which “turn subjective experiences
into emotional experiences” [14 p. 146]. Once triggered, an
emotional response can then involve a resetting of procedural
priorities and the regulation of the speed of task execution
(which can be set to 0 as in case of a freeze-reaction) thus
producing a cascade of organismic changes. It goes without
saying that the sensitivity thresholds, as well as ceiling values
will vary from one person to the other depending on genetic,
physiological, contextual and social factors.
The boundary between an emotional and non-emotional
response is thus set by each individual's critical-range values
in a three-dimensional affective space. As long as the
person’s monitoring system reports values of valence, arousal
and power within his/her range of normal variation, the state
will not be subjectively experienced as emotional.
Behaviourally speaking adaptive action/reaction patterns have
a range of foreseeable fluctuations which represent the shock
absorption capacity of a particular system without it being
endangered in its structural integrity. In other words the state
of the organism - related to each moment's psycho-
physiological configuration - is always affectively coloured.
We suggest that the concept of emotion can thus be
defined as a cognitively dominant body-mind configuration
presenting one or more out-of-normal-range values on at least
one of the three dimensions: valence, arousal and power.
There is reason to assume that the shifts in the overall
configuration are non-linear.
We believe that a unified model of cognition and emotion
can constitute a suitable conceptual framework for studying
acoustic correlates of affective states. It can explain why

emotions, attitudes, moods, personal stances and even
personality traits can be considered as different types of
affective states or processes. It is because the three elementary
affective dimensions are inherent to all of them.
The advantage of using such a three-dimensional
approach over a categorical labelling approach lies in the fact
that one can establish a continuity from non emotional to
emotional states and processes. This is in particular
appropriate for studying natural corpora.
2.2. A place for affective prosody in a unified model of
cognition and emotion
Speech communication, just as any other interactive
behaviour, necessarily reflects the three dimensions of the
affective space. These dimensions are then encoded in many
aspects of speech and in prosody in particular. Prosody
partakes at the same time of the physical, the physiological,
the mental, the individual and the social. It reflects the
configuration of the speaker's affective space in two ways: the
underlying configuration extending over a longer period of
time (moods, stable attitudes, and personality traits) and each
moment's on-line configuration (emotions, interpersonal
stances and communication strategies). In natural settings the
on-line configuration is influenced by ever changing
discursive context including the speaker's and the receiver's
own non-verbal reactions, the discourse content and the
conversational interaction pattern itself.
We are now faced with a number of fundamental
questions:
What exactly does affective prosody indicate? What
meaning does the receiver attribute to prosodic variations?
What is the semiological status of vocal affect signalling? Is
affective prosody a symptom, a symbol or a conventional
sign? This poses a further -seemingly terminological- problem
of expression vs. communication of affect whereby
expression = external manifestation of an inner state
(Darwin’s meaning), while communication implies intentional
signalling [25].
To answer these questions let us consider the following
semiological distinctions regarding basic types of semiotic
entities involved in the mechanism of indication.
1. Spontaneous indicators [25] are double-faced entities
where the link between the signifier and the signified is
naturally given. It is said to be motivated [21 pp. 134-
135] because the link reflects relations such as: spatial or
temporal contiguity, causality, implication or pars pro
toto relationship. Example: footprints in snow,
temperature as a symptom of illness. In other words,
spontaneous indicators have an informative value without
having been produced to this purpose.
2. Falsely spontaneous indicators [3] are those that are
purposely produced in order to appear as natural or
spontaneous. Example: a foreign accent produced by a
native speaker wanting to appear as a foreigner.
3. Intentional indicators are double-faced entities which are
produced in order to provide information other than their
own existence. These include:
∑ symbols, where an originally natural link between the
signifier and the signified has been conventionalised for
purposes of communication. Example: a picture of a
snake symbolising a pharmacy. In the course of the
history, a symbol may loose its previously motivated
nature.
∑ icons, where the link between the signifier and the
signified is motivated by topological similarity between
the two [35]. Examples of icons: maps, images, diagrams,
metaphors
∑ signs, where the relationship between the signifier and the
signified has been arbitrarily established and
conventionalised for purposes of communication [28]
Example: words of a language.
We believe that the semiological status of affective
prosody can be that of: a spontaneous indicator (a symptom),
a falsely spontaneous indicator, or a symbol but never a sign
as defined above. The non arbitrary nature of prosodic
indicators is supported by cross-cultural studies showing that
the subjects accurately infer emotional states from acoustic
cues produced in other cultures and languages [6].
Parallelisms in animal affect signalling is of interest here as
well.
So, what exactly does vocal expression indicate?
Scherer's model of vocal-affect-signalling provides
precise predictions for phonetic and macro-prosodic changes
related to the outcome of each of the components of his model
of appraisal [31]. It therefore follows that, emotionally
induced phonetic and prosodic variations primarily have a
status of spontaneous, natural indicators or symptoms of
physiological reactions automatically triggered by cognitive
appraisal outcomes. They reflect the “push” force in affect
signalling [34]. An example would be low pitch variability,
and low overall intensity usually found in sadness. We believe
that in the case of intense primary emotions (anger, fear, joy,
sadness and disgust) the push effects are dominant and the
speaker may have little freedom in influencing his vocal
expression. A special case may be made for highly distressful
situations where vocal characteristics are intentionally
modified for purposes of better transmission over a distance
and as cues for the sender's localisation.
By contrast the “pull” force in vocal affect signalling is
related to norms or expectations imposed by the physical or
social environment which require production of specific
acoustic features allowing the sender to achieve a particular
effect. The pull force may restrict or enhance some of the
surface features of the push effects. It therefore follows that,
when push and pull factors are blended, vocal signals have a
two-fold status of symptoms and symbols. When pull effects
dominate, vocal signals may have a status of falsely
spontaneous indicators or symbols. An example would be the
use more highly-pitched voices in signalling submission and
appeasement in friendly or submissive encounters.
The semiological status of affect signalling can thus vary
from symptom to a conventionalised or ritualised symbol.
This also raises the issue of speaking styles. Is any individual
speaking pattern a style? For Prieto a style is related to the
possibility of choosing a means (in this case a vocal pattern)
for achieving an aim [26 p.99] . It therefore follows that if
there is only one possible means of achieving an aim, the
subject cannot have a style. Affectively marked vocal
expression will be a style as long as it is a result of the
speaker's choice. In this context Prieto distinguishes between
reactions and behaviours. The latter (speech behaviour
included) involve engaging in action by one's own decision in
a situation of choice. For Prieto, it is the power to choose how
to act, (i.e. the possibility to have a style) that allows the
subject (the speaker) to build his own Self. Prosodic choice
thus appears to be a psychologically important factor.

Another refinement may be necessary: the distinction
between the unconscious purely physiologically driven push
factors and the more psychologically driven push factors. In
the latter case the choice of expressive style may be regarded
as expression of the speaker's identity. An example would be
persistent usage of speech patterns related to one’s own native
regional dialect.
The relationship between push and pull factors varies
according to context. Is highly probable that blends of push
and pull forces are more frequent than pure dominance of one
or the other. In the Table 1. we present the hypothesised
relationships between the two forces for five categories
cognitive-affective configurations with corresponding values
of affective space tags and the semiological status of vocal
signals. The five cognitive-affective configurations
correspond to the five different types of affective states and
traits presented by Scherer as examples of design feature
delimitation of different affective states. [33].
Table 1: Features of five cognitive-affective
configurations
Affective-
cognitive
configuration
Affective
space
tags' values
Push-Pull
relationship
in speech
behaviour
Semiologi
cal status
of vocal
signalling
Emotions in critical
range
push
dominant
mainly
symptom
(possibly
symbol)
Moods
within normal
to border line
range
push
dominant or
blended
mainly
symptom
Attitudes within normal
range
push
dominant or
blended
mainly
symptom
(possibly
symbol)
Interpersonal
stances
within normal
range
pull
dominant
mainly
symbol
Personality
traits
within normal
range
push
dominant
symptom
In comparison with other interactive behaviours, speech
communication is specific in that it implies the existence of a
receiver able to decode the signal and attribute a meaning to
the speaker's utterance by taking into account the context in its
physical, social and discursive dimensions. Because of the
intrinsically natural relationship between prosody, and the
speaker's cognitive-affective and physical state, we believe
that affective prosody can be regarded as a parallel natural
code which provides contextual information thus helping the
receiver to disambiguate the meaning of the utterance. The
interpretation of the speech signal is mainly probabilistic, and
driven by physical and social context including stereotypical
representations of normal vs. affectively marked speech
patterns. The greater the mismatch between the expected and
the actual, the more likely it is that it triggers the search for
interpretation or connotative meaning attribution.
2.3. Prosody as a symptom of arousal dimension: example
of flattening of vocal arousal due to autonomic neural
lesions in diabetic patients
2.3.1. Hypotheses and method
Given that basic autonomic responsivity influences emotional
reactions, we hypothesised that the subjects with autonomic
lesions would display diminished emotional reactions resulting
in a diminished vocal arousal and consequently a flattening of
vocal differentiation between high and low arousal emotional
states. The subjects were 40 diabetic patients. The autonomic
responsivity was assessed by means of two standard
autonomic function tests based on the quantification of heart
rate variability (HRV) [36]. These provided a HRV index
indicating each patient's level of autonomic responsivity. It
was also hypothesised that low autonomic responsivity would
be related to a diminished subjective experience of emotions
The latter was assessed through verbal self report.
Emotional states were induced through verbal recall of
personal emotional experiences of anger, joy, and sadness. At
the end of each recall the patients were asked to pronounce,
on a mood congruent tone, the sentence: “ALORS TU
ACCEPTES CETTE AFFAIRE” (“So you accept the deal”).
The sentence was presented in capital writing without
punctuation so as not to suggest any tone of voice. The
patients were then asked to what degree they had subjectively
felt the emotion described during their recall. The results were
coded on a 4-point scale, ranging from not at all to very
much. The patients' voices were acoustically analysed for
standard vocal parameters related to F0, energy and delivery
rate. Upon inspection of partial correlations with HRV index
(controlling for age, anxiety state and extroversion), three
vocal parameters appeared as significantly correlated with
HRV index. These were: F0 max/min ratio, voiced energy
range and the rate of delivery. To obtain an index of the
overall degree of vocal arousal related to each condition
(anger, joy, sadness), we calculated a summary score
composed of z values of the three above mentioned vocal
parameters. The summary score was named Vocal arousal
index. As we expected the subjects with high HRV index to
exhibit higher vocal arousal in anger than in sadness, we then
calculated the delta between the Vocal arousal index in anger
and that in sadness. Each subject was thus characterised by
his or her Vocal differential index reflecting the degree of
his/her vocal differentiation between anger and sadness.
2.3.2. Results
As we expected the Vocal differential index to be correlated
with autonomic responsiveness, we performed liner multiple
regressions (stepwise method) with Vocal differential index as
dependent variable and HRV index, age, sex, state anxiety,
and extraversion scores as independent variables. The results
of the regressions showed a highly significant effect for HRV
index (T = 7.19; p < 0.001) and a much lesser effect for state
anxiety (T = -2.052; p = 0.05). The HRV index alone
explained 58% of data variance with the multiple R = .79. The
subjects with higher degree of autonomic responsivity
displayed higher prosodic differentiation of emotions. None of
the other variables contributed significantly.
From the above results we conclude that, in our subjects,
poor prosodic differentiation between anger and sadness
could be interpreted as a symptom of poor autonomic

responsivity and marginally as a symptom of the state
anxiety.
To test our second hypothesis regarding the relation
between autonomic responsiveness and the degree of self
reported subjective feeling, we performed Mann-Whitney U
tests on 2 groups obtained by median split on HRV index. The
results showed significant differences in the degree of felt
sadness (Z = - 3.3; P = .0009), and anger (Z = - 2.4; P = .02).
The groups with higher HRV reported higher degree of
subjective feeling for both sadness and anger than those with
lower HRV. For details and discussion, see [37].
These results lend support to the hypothesis that subjects
with poorer autonomic responsivity experience flattening of
emotional reactions on both the level of vocal expression, and
the subjective feeling.
2.4. Prosody as a symptom of power dimension: a pilot
study of adaptive vs. maladaptive coping style in breast
cancer patients
2.4.1. Hypotheses and method
Active coping style has been defined as “tonic readiness to act
upon an event”, while passive coping was described as
“waiting for and enduring and attending to stimuli” [18]. It
was hypothesised that coping power, as attitudinal dimension
related to the appraised power, could be reflected in the
prosody of spontaneous speech. Ten breast cancer patients
were interviewed by trained clinicians who rated their
adjustment to the disease and treatment as adaptive vs.
maladaptive. The definition of adaptive coping style
corresponded to Obrist's concept of active coping. The
interview was semi-structured and adapted from an interview
evaluating mental adjustment in patients with early breast
cancer [17]. Samples of the patients’ voice recordings were
taken at various moments of the interview that is: a moment of
maximum vocal arousal, one of minimum vocal arousal as
well as in passages related to specific topics in the course of
the interview. They were acoustically analysed for the
following voice parameters: mean F0, F0 range expressed as
max/min ratio, F0 coefficient of variation, mean intensity,
intensity range expressed as max/min dB ratio, and the rate of
delivery. For each of these parameters we measured the
difference between the values for high and low arousal
conditions.
2.4.2. Results
Of particular interest is a remarkable higher difference
between F0 range in high and low arousal conditions (DF0
range) in cases with adaptive (median = 55.2, 90% CI = 46.0-
60.4) versus those with maladaptive adjustment (median =
35.2, 90% CI = 8.0-42.4), with Wilcoxon-Mann-Whitney U =
25.0; P = 0.009. In spite of our small sample, we believe that
the results indicate a link between prosodic features and the
power dimension inherent to attitudes related to coping styles
and mental adjustment [1].
3. Discussion
Although the concept of three dimensional affective space is
not novel [9] [16] [20], we believe that so far it has not been
integrated into a unified model of cognition and affect
whereby the very process of cognition is functionally related
to a three-dimensional affective space, which in turn
determines the subject’s subjective feeling as well as his
actions involved in adaptive behaviour. We are also aware of
the fact that, space not permitting, the model has been
presented in a schematic way and further sophistication is
much needed. We feel that research in affective prosody can
benefit from a unified model as it offers a continuity from the
states and processes traditionally regarded as cognitive (doubt,
certainty) to emotionally coloured interpersonal stances
(friendly), and to full fledged emotional experiences. For an
example of application of valence, power and arousal as
dimensions of emotional meaning see the work of Cécile
Pereira [20].
4. References
[1] Bernhard, J.; Zei Pollermann, B., 2002. Emotional
expression as an indicator of mental adjustment: Voice
analysis in patients with early breast cancer. Submitted
[2] Bertalanffy, von L., 1968. General System Theory. New
York: George Brazilier.
[3] Buyssens, E., 1943. Les langages et le discours.
Bruxelles: Office de publicité.
[4] Cacioppo, J.T.; Klein, D.J.; Bentson, G.G; Hatfield, E.,
1993. The psychophysiology of emotions. In Handbook
of Emotions, M. Lewis and J.M. Haviland (eds.),. New
York: Guildford, 119-142
[5] Chapman, C.R.; Nakamura, 2001. The affective
dimension of pain: mechanisms and implications. In
Emotions, Qualia, and Conscoiusness, A. Kaszniak (ed.).
Singapore; London: World Scientific, 181-210.
[6] Clynes, M.; Netttheim, M., 1982. The living quality of
music: Neurobiologic basis of communicating feeling. In
Music, Mind, and Brain, M. Clynes (ed.). New York:
Plenum, 47-82.
[7] Damasio, A.R., 1994. Descartes' Error. Emotion, Reason
and the Human Brain. New York: G.P. Putnam's Sons.
[8] Davidson, R.J., 1992. Emotion and Affective style:
Hemispheric differences. Psychological Science, 3, 39-
43.
[9] Davitz, J.R., 1964. Auditory correlates of vocal
expression of emotional feeling. In The communication of
emotional meaning, J.R. Davitz (ed.). New York:
McGraw-Hill, 101-112.
[10] Duffy, E., 1941. An Explanation of “emotional”
phenomena without the use of the concept of emotion.
The Journal of General Psychology, 25, 283-293.
[11] Ekman, P.; Davidson, R.J., 1994 (eds). The nature of
emotion. Fundamental questions. New York - Oxford:
Oxford University Press.
[12] Lane, R. D.; Nadel, L.; Allen, John J.B.; Kaszniak A.W.,
2000. The Study of Emotion from the Perpective of
Cognitive Neuroscience. In Cognitive Neuroscience of
Emotion. R.D. Lane and Nadel, L. (eds). New York -
Oxford: Oxford University Press, 3-12.
[13] Lazarus, R.S.; Coyne, J.C.; Folkman, S., 1984. Cognition,
emotion and motivation. The doctoring of Humpty-
Dumpty In Approaches to emotion, K.R. Scherer & P.
Ekman (eds.). Hillsdale, N.J.: Lawrence Erlbaum
Associates Inc., 221-237.
[14] Ledoux, J., 2000. Cognitive Emotional Interactions:
Listen to the Brain. In Cognitive Neuroscience of
Emotion. R.D. Lane and L. Nadel (eds.). New York -
Oxford: Oxford University Press, 129-156.

[15] Leventhal, H.; Scherer, K. 1987. The Relationship to
Emotion to Cognition: A Functional Approach to a
Semantic Controversy. Cognition and Emotion, 1, 3-28.
[16] Mehrabian, A.; Russel, J., 1974. An approach to
environmental psychology. Cambridge : MIT Press.
[17] Morris, T.; Blake, S.; Buckley, M., 1985. Development of
a method for rating cognitive responses to a diagnosis of
cancer. Soc. Sci. Med. 20, 795-802.
[18] Obrist, P.A., 1981. Cardiovascular psychophysiology: a
perspective. New York: Plenum.
[19] Ohman, A., 1999. Distinguishing unconscious from
conscious emotional processes: Methodological
considerations and theoretic implications. In Handbook
of Cognition and Emotion, T. Dalgleish and M. Power,
(eds.). Chichester, UK: Wiley, 321-352.
[20] Pereira, C., 2000. ISCA Workshop on Speech and
Emotion: A conceptual framework for research.
http://www.qub.ac.uk/en/isca/proceedings/index.html.
Proceedings Online.
[21] Piaget, J., 1967. La psychologie de l'intelligence. Paris:
Armand Colin.
[22] Piaget, J., 1970. Epistémologie génétique. Paris: Presses
Universitaires de France.
[23] Piaget, J., 1970. Problèmes généraux de la recherche
interdisciplinaire et mécanismes communs. In: Tendances
principales de la recherche dans les sciences sociales et
humaines, première partie : Sciences sociales. Paris ; La
Haye : Mouton ; Paris: Unesco, cop., 559-628.
[24] Piaget, J., 1981 / 1954. Intelligence and Affectivity: their
Relationship During Child Development. Palo Alto:
Annual Reviews.
[25] Prieto L.J., 1975. Pertinence et pratique. Paris: Editions
de Minuit.
[26] Prieto L.J., 1975. Etudes de linguistique et de sémiologie
générales. Genève: Droz.
[27] Roseman, I.J., 1991. Appraisal determinants of discrete
emotions. Cognition and Emotion, 5, 161-200.
[28] Saussure, F. de, 1972. Cours de linguistique générale.
Paris: Payot.
[29] Schachter, S.; Singer, J.E., 1962. Cognitive, social and
physiological determinants of emotional state.
Psychological Review, 69, 379-399.
[30] Scherer, K.R. 1984. Emotion as a multicomponent
process: A model and some cross cultural data. In Review
of personality and social psychology, P. Shaver (ed.).
Vol. 5. Beverly Hills, CA: Sage, pp. 37-63.
[31] Scherer, K.R. 1986. Vocal affect expression: A review
and a model for future research. Psychological Bulletin,
99, 143-165.
[32] Scherer, K.R., 1993. Neuroscience Projections to Current
Debates in Emotion Psychology. Cognition and Emotion,
7, (1), 1-41.
[33] Scherer, K.R., 2000. Psychological models of emotion. In
The neuropsychology of emotion, J. Borod (ed.).
Oxford/New York: Oxford University Press, 137-162.
[34] Scherer, U.; Helfrish, H.; Scherer, K.R., 1980. Internal
push or external pull? Determinants of paralinguistic
behaviour. In Language: social psychological
perspectives, H. Giles; P. Robinson and P. Smith (eds.).
Oxford - New York: Pergamon, 279-282.
[35] Sebeok, T.A., 1976. Contributions to the Doctrine of
Signs. Bloomington: Indiana University, with Lisse: The
Peter de Ridder Press.
[36] Vita, G.; Princi, P.; Calabro, R.; Toscano, A.; Manna, L.;
Messina, C., 1986. Cardiovascular Reflex Tests. Journal
of the Neurological Sciences, 75, 263-274.
[37] Zei, B.; Archinard, M., 2001: The role of autonomic
balance in experiencing emotions. In Emotions, qualia,
and consciousness, A. Kaszniak (ed.). Singapore, World
Scientific Publishing, 278-284.