What do dimensional approaches to the structure of affect look like? - Chapter 5

What is a dimensional view of affect?

Many emotion scientists view emotions and moods as reflections of general dimensions of experiences, rather than as discrete categories. Therefore, this concerns how the world is experienced, this is often examined by self-report. An assumption is that emotions are largely defined by the verbal labels that we associate with them. The problem with this is that we do not know whether a certain concept really refers to a biological or psychological phenomenon. So, it may be that our experience of emotions is determined by our language and culture, which would mean that emotions are learned or socially constructed, rather than something natural.

What do we mean by dimensions?

According to the dimension approach, the human experience of emotions (i.e. feeling) is the most important phenomenon that must be explained. This emphasis on feeling leads to the hypothesis that the world is experienced along broad dimensions, rather than categories of different emotions. Much research has ultimately revealed two important dimensions along which affect can be described: arousal (activation) and valence (valence, positive / negative).

Arousal describes the amount of energy that we feel is available. This can range from deep sleep and drowsiness on one side of the scale to hyperactivity on the other. Arousal changes during the day, depending on environment, time, activities, weather, etc. In addition, it depends on personality differences. These variations are felt as changes in the so-called core affect , and form an important dimension of our subjective experience of affective state. Valence or pleasure refers to whether the state we are in is experienced as positive or negative, pleasant or unpleasant.

Arousal and valence seem to be universal dimensions in human experiences. This two are the most salient dimensions, and it is very difficult to add a third dimension that is just as distinctive. It is important that arousal is not a separate phenomenon in the brain, but rather that it represents variations in the valence dimension.

What are the criteria for identifying general dimensions of affect?

The idea that we experience a mix of broad dimensions instead of discrete emotions seems somewhat counter-intuitive, because our perception of discrete emotions is very strong. But when you imagine a situation where you were very scared and one in which you were very angry and trying to describe it, it is difficult to go beyond a description of arousal and valence. The only thing that distinguishes the two is perception and interpretation of the context.

Overview of empirical evidence for dimensions subjective report

Description experience sampling measurements of affective experience

In the Descriptive Experience Sampling procedure, large numbers of people are tested about different situations and they are asked how they feel in each situation. Some describe their emotions in more specific terms of emotions, but all participants also describe their feelings in response to valence. This indicates that valence is a more fundamental characteristic of affect than discrete emotion terms.

Emotional granularity refers to whether someone uses discrete emotion terms in a very precise way (high on emotional granularity), or uses them in a more global way (low on emotional granularity). This indicates that people differ in how good they are in describing their internal affective state. It seems that anger, sadness, fear, etc., are often not experienced as clearly distinct states. Therefore, these states must be broken into smaller parts that relate to valence and arousal.

Questionnaire measurements of affective experience

When self-report data of the experience of emotions is placed in a graph, comparing valence and arousal, a circumplex structure is usually created. This includes the basic emotions in a certain quadrant of the model. This suggests that affect can best be seen as variations based on a number of dimensions, rather than as discrete categories. Every emotion is therefore a combination or mix of these dimensions.

Scores on a dimension do not have to be completely dependent on each other. In extreme cases this is true: when someone experiences extreme positive affect, he usually feels very little negative affect. However, if someone is relatively low on negative affect (he is fairly relaxed), that does not necessarily mean that he feels happy. This shows that positive and negative affect are surprisingly independent of each other. Scores can be anywhere on the continuum. The two-dimensional model then has two axes, one for positive affect and one for negative affect, on which the scores can be plotted.

It is also possible to make a model with the dimensions of activation: tension and energy. All of these models seem very different, but it appears that they all actually describe the same affective space. When valence and activation are used to describe the results of these studies (instead of the original dimensions), the explained variance was still very high.

Physiological specificity

The idea here is that approach or withdrawal mechanisms are an explanation for the valence dimension. These mechanisms are mutually inhibiting, so when one is activated, the other is suppressed. Arousal is not seen here as a third system, but as the metabolic and neural activation of either approximation, withdrawal, or the co-activation of both together. Therefore, the approach and withdrawal mechanisms are seen as the neuro anatomical basis of both valence and arousal effects. If this is true, we should find physiological responses that are triggered by increasing arousal, but not influenced by valence (or vice versa).

Facial muscle action

Important muscles in the face include the corrugator muscles (responsible for contracting the eyebrows, frowning) and the zygomatic muscle (responsible for smiling at joy). Activation of these muscles therefore gives an indication of positive or negative affect (valence). This is about the degree of activation: there is always activation, but the more positive the stimulus is experienced, the less activation can be seen in the corrugator muscle. Logically, the opposite effect occurs with the zygomatic muscle. There is a difference in gender: men meet the expected outcomes much less often. It seems that women are more expressive with their faces when they experience positive or negative affect

Heart rate and skin conductance

Variations in heart rate are used as an index for the response to emotionally charged stimuli. The default pattern is a decrease in heart rate when a photo first appears, then an acceleration and finally a decrease. The valence of the photo influences the extent to which these changes occur: unpleasant stimuli produce the greatest degree of initial deceleration and acceleration (in both the first phase and the last), while pleasant stimuli produce greater peak acceleration. Heart rate is therefore a measurement for valence.

A more reliable method is to measure skin conduction. The degree of skin conduction increases when arousal increases, this is independent of valence. The skin guidance therefore increases when strongly emotionally charged images are shown, both positive and negative. The fact that the dimensions are so clearly separated indicates that this is indeed a matter of two separate factors.

Startle response

Measuring a startle response to an emotional stimulus appears to be a good way to test whether the activation of the withdrawal mechanism (and deactivation of the approximation mechanism) is influenced by the overall level of activation (arousal). A startle response is a protective reflex that interrupts current behavior, so that a potential threat can receive attention. According to the motivational priming hypothesis, this shock reaction should be stronger when the withdrawal mechanism is already activated (for example, when watching a horror movie).

The startle reflex is measured in humans by measuring the eye reflex (where someone blinks with an eye when he is shocked). When shock-conditioned stimuli are processed, the startle reflex is stronger as a result of a shock stimulus (for example, a loud bang). This is consistent with the before mentioned motivational priming hypothesis. Because the activation of the withdrawal mechanism is accompanied by the suppression of the approximation mechanism, the startle reflex during the processing of a pleasant stimulus has a lower intensity. The higher the arousal, the stronger the reflex with positive stimuli and the less strong the reflex with positive stimuli.

Dillon and LaBar found that conscious attempts to increase or suppress the emotional response to an image had similar effects on the fright response (independent of valence). This is inconsistent with the motivational priming hypothesis, but indicates that arousal is more important for regulating the shock reaction than valence.

How could neural circuits be identified?

Identifying neural circuits for valence and arousal is difficult. If they are two separate dimensions, we would expect certain areas of the brain to be activated in response to arousal, but not to valence. This is even more complicated with valence, because it is not clear whether valence is one dimension with two extremes, or whether positive affect and negative affect are each a separate, independent dimension.

A model in which positive affect and negative affect vary in intensity seems logical, because the intensity often has an influence on how pleasant or unpleasant something is experienced. This model thus suggests that valence increases on the basis of arousal, from the most neutral to the most intense, regardless of whether this is positive or negative.

EEG studies of the neural factors of valence

The valence asymmetry model suggests that the right and left prefrontal cortex play different roles in how we perceive and experience positive and negative emotions. Positive emotions seem to cause greater activity in the left PFC, negative emotions in the right PFC. It is not always the case that negative emotions are related to withdrawal. Anger, for example, is more focused on approach. So, it would be too easy to say that positive and negative emotions are the same as approach related and withdrawal related emotions. The question therefore is whether the activity in the left or right PFC depends on valence or on the approach and withdrawal mechanisms.

What is the functional neuroanatomy of valence?

Two meta-analyzes of a large number of studies on the neuroanatomy of valence found very inconsistent results. One analysis showed that approach-related emotions were associated with greater activation in the left PFC. However, no differences were found between the two hemispheres for the withdrawal mechanisms. Another analysis found that positive emotions were associated with greater activation of the left lateral PFC and basal ganglia, while negative emotions were associated with increased activation of the insula. Here too, left PFC activation was related to approach emotions, along with the activation of the medial PFC. With withdrawal emotions, the amygdala, left medial PFC, anterior cingulate, basal ganglia, left insula and left fusiform, and superior occipital cortices were related.

Another study found the following: regardless of whether an image was negative or positive, if the image was emotionally charged, there was an increased activation of the vmPFC. This relationship is therefore completely independent of valence. If the images also had to be classified on the basis of valence, there was activity in the dlPFC, which therefore seems to influence emotional assessment. The vmPFC therefore probably relates to the affective significance (both positive and negative) of an emotional stimulus, while the dlPFC relates to the evaluative aspect of valence.

What is the functional neuroanatomy of arousal and valence?

Recent fMRI studies have shown that the activation of the amygdala is specifically correlated with emotional arousal, while the amygdala does not respond to differences in valence. Validity was associated with an increase in activity in the orbitofrontal cortex (OFC), this area again did not respond to differences in intensity (arousal). Lewis even found that different parts of the OFC were concerned with positive or negative valence. Positive valence was linked to the right lateral OFC and the anterior insula, while negative valence was related to the posterior insula, the ACC and the right and right medial OFC.

Lewis also found that large parts of the brain were activated by the conjunction of valence and arousal, allowing the two to be integrated into a single representation. This valence-specific response to arousal supports the idea that emotional content influences the way we respond to arousal.

The neuroscience of reward and punishment

Rolls' schedule classifies emotions and terms of rewarding effects. His model describes on the vertical axis the emotions associated with the award of a reward (up) or punishment (down). On the horizontal axis, it describes the emotions associated with non-delivery of an expected reward (to the left) or the non-delivery of an expected punishment (to the right). However, this is not about dimensions, because the parameters of the model are not completely independent of each other.

The OFC seems to play a major role in coding stimuli related to reward or punishment. For example, this is activated when someone eats their favorite food, or even thinks about it. The scores that people indicated how pleasant the food was slowly decreased as they ate more and became saturated and this was also reflected in reduced activity in the OFC. Here too, different parts of the OFC were active, depending on whether a score had to be given on how enjoyable the food was or how unpleasant. The medial OFC plays a role in coding rewarding stimuli, the lateral OFC in evaluating punitive stimuli. The nucleus accumbens (NAcc) ensures that dopamine is released. This is associated with experiencing positive affect and reward. Both the OFC and the NAcc are therefore important areas for coding reward and punishment.

A reward is not just a simple matter of dopamine release in the NAcc. There are three processes that are necessary for a stimulus to acquire rewarding properties. First, the individual must be motivated to act and learn. Secondly, he must learn about the relationship among stimuli and the consequences of actions relating to these stimuli. Finally, the consumption of the reward must cause hedonic consequences, in the form of pleasure. Each of these three processes consists of both explicit and implicit components. The explicit (desire, expectation, pleasure) are consciously experienced, the implicit (habits, salience of the incentive) not. According to Berridge and Robinson, variations in both motivational and emotional components can occur unconsciously, even if the subjective scores that people give to their mood and feelings do not change.

Agents that influence the level of dopamine release do not seem to have any influence on the subjectively experienced pleasure due to drugs or eating. Opioids, on the other hand, do have an influence on subjective pleasure: an injection of opioids leads to more expressions of pleasure, while blocking the opioid receptors reduces the reward value. Opioids therefore appear to be involved in liking and increasing pleasure due to sensory stimuli.

The incentive salience of a stimuli, also called wanting (wanting to have), is strongly influenced by dopamine, but also depends on other brain structures such as the connections between the NAcc and the amygdala and parts of the cortex. This explains the fact that the manipulation of dopamine systems has strong effects on motivated behavior but not on the degree of 'liking' of the reward.

Cognitive appraisals

Because people also indicate that they feel a number of clearly distinguishable emotions, dimensional theories must also show that cognitive appraisal can lead to the impression of discrete emotions.

Are valence and excitement the dominant dimensions of evaluation?

The first step in the stimulation evaluation process is the evaluation of valence; is something good or bad? On a subjective level there are great similarities between different emotions. Although anger and fear can be distinguished from each other by facial expression, they are experienced in the same way (unpleasant and exciting). To properly map out these similarities, a dimensional approach is needed. Such an approach can explain why people move so easily from one emotion to another.

Smith and Ellsworth developed a model consisting of six dimensions that allowed differentiation between fifteen different emotions. For most of these emotions, a unique pattern of cognitive evaluations could be identified. The evaluations of valence and arousal were also included in this list. A strong relationship was found between the cognitive interpretation of an event and the emotional response to it.

In general, research into cognitive evaluations does not agree with the approach that there are only two dimensions. It is assumed that people consider very many dimensions in the composition of their emotional experience.

Does cognitive interpretation of affective state lead to distinct emotions?

James' theory that physiological arousal is the cause of emotions has a number of shortcomings. In this way the same arousal arises even when we exercise hard, but it is unlikely that we will feel the same emotion. In addition, physiological changes are too slow to precede a conscious experience of emotion. Another problem is that people who experience very different emotions nevertheless experience almost the same physiological arousal. Cannon therefore argued that emotion arises when the thalamus simultaneously sends signals to the cortex (for the conscious experience of emotion) and the autonomic nervous system.

Schachter and Singer - two-factor model

Schachter and Singer stated that experiencing emotion depends on two factors: physical arousal and cognitive interpretation of that arousal. So if you experience a physiological excitement, look for an explanation in your environment. This is partly in agreement with James, but also explains how it is possible that the same arousal results in a different emotion. So an emotional experience has two parts: a physiological element and an (externally focused) cognitive appraisal element.

This theory was confirmed in an experiment in which subjects were given adrenaline or a placebo. One part was well informed about the side effects (arousal with the adrenaline and no side effects with the placebo) and another part received the reverse (so wrong) information. It turned out that when someone thought the adrenaline had no side effects, he blamed the arousal on another person in the room. When this person behaved happily, the subject also felt happy, but when this person behaved angry, the subject also felt angry. The arousal is therefore combined with an appraisal of the situation: "I am excited, so that will be joy / anger because this other person behaves that way."

What are common misattributions of arousal?

Much research shows that physical arousal colors our subjective experience of the world. For example, there was an investigation where men had to cross a river over a low stone bridge, or over a high, wobbling, narrow bridge. Then a female researcher came to them and gave them a questionnaire. However, they also got her phone number asking if they wanted to call her after the investigation. The men who had walked over the wobbly bridge (and were therefore more aroused) called more often. They had wrongly attributed their arousal caused by the bridge: they thought it was because they were attracted to the researcher.

The NAcc is anatomically connected to a number of areas in the PFC, but also to a number of subcortical areas such as the amygdala. It is therefore possible that the NAcc plays an important role in determining the affective significance and saliency of a stimulus, and not specifically arousal or valence.

The conceptual act model

The conceptual act model states that discrete emotions are an illusion created by a cognitive categorization process called core affect. The hypothesis here is that what we perceive and experience as individual emotions actually comes from how we perceive and categorize changes in a more general core affective system.

In other words: our background moods (core affect) are experienced along the general dimensions of valence and arousal. These states are subdivided into discrete emotion categories based on a cognitive appraisal of the current context.

What is the difference between discrete and dimensional approaches?

Researchers who look at discrete emotions often focus on the neural or physiological basis of emotions or moods. Some even think that the subjective experience of emotions is irrelevant for understanding the emotion. On the other hand, researchers who start from a dimensional approach focus on subjective experience of feelings. In addition, dimensional research is more often done with humans, while discrete emotion supporters do research with rats more often.

The challenge of research into emotion is to compare these two different research traditions and see if the empirical evidence of both can be integrated into one explanatory model of affect. Both traditions provide information from different levels of analysis (subjective reporting versus neural activation). This data could therefore be merged into a comprehensive model.

Finally, it is unfortunate that individual differences in how people react in emotional situations and how they perceive and interpret these situations are almost ignored. These are influential aspects in the structure of affect.