Not in the Mood: Affective Forecasting and Cognitive Architecture

Andrea Blomqvist is a PhD candidate at the University of Sheffield, where she is writing her thesis on affective forecasting. Her work is interdisciplinary, sitting between philosophy, psychology, and cognitive neuroscience. Her other research interests unsurprisingly include imagination, memory, and affect. When she's not working at a desk, she enjoys the great outdoors - especially running and gardening.

Andrea Blomqvist is a PhD candidate at the University of Sheffield, where she is writing her thesis on affective forecasting. Her work is interdisciplinary, sitting between philosophy, psychology, and cognitive neuroscience. Her other research interests unsurprisingly include imagination, memory, and affect. When she's not working at a desk, she enjoys the great outdoors - especially running and gardening.

A post by Andrea Blomqvist.

Why is it that some things are easier to imagine than others? A substantial part of the answer can be formulated by looking at the cognitive architecture of the human mind (i.e. the structure of the mind), which is what I will do in this post. Here, I will explore how the cognitive architecture of our affective system influences the ‘affective forecasting’ system - the capacity we use when we try to accurately imagine (or forecast) our future moods and emotions in order to make decisions. When we affectively forecast, we do something more than just trying to imagine the phenomenal character of a mood or emotion; we try to imagine the phenomenal character that we actually will experience in a future scenario. For example, to decide whether or not to move to a new city, you can use imagination to figure out how you are going to feel when you are there; or, as in L.A. Paul’s example, to decide whether or not you want a child, you can try to imagine what having a child will be like and how that will make you feel (Paul, 2004). We can also use it in more mundane cases, like imagining how you will feel if you do badly in an exam, or if you are rejected by a date (Wilson and Gilbert, 2000; Levine et al., 2012).

Experiments on affective forecasting have suggested that we are not flawless at imagining how we are going to feel in the future (Gilbert et al., 1998; Wilson et al., 2000). An initial problem is that we need to imagine the right kind of scenario. If I imagine how going on holiday will make me feel, I might imagine staying in a luxury hotel by the seaside. Consequently, I may believe that I will feel happy and relaxed during my holiday. However, my limited budget means that I could never afford a luxury hotel, and the cheap hotel I actually stay in does not leave me even nearly as happy and relaxed as I imagined. Here, I have failed to imagine the right kind of scenario, and my affective forecast is thus not accurate. Still, as we will see, even when I do succeed in imagining the right scenario, there are further difficulties due to the structure of our cognitive architecture.

I will argue that it is harder to correctly forecast moods compared to emotions, because moods, unlike emotions, are not about anything. In fact, I will suggest that most of the time when we try to forecast a mood, we actually fall back on forecasting an emotion instead. Let me explain further by first distinguishing between the cognitive architecture of emotions and moods.

Elaborating on James (1884) and Prinz (2004), Baralssina and Newen (2014) have put forward the so-called Impure Somatic Theory of Emotion, whose main tenet is that an emotion is constituted by an integration of an interoceptive state (a representation of one’s own bodily state) and a representation of an external object. Let us consider fear of a snake to illustrate this. As a neo-Jamesian view, it holds that my heart does not race because I feel fear; I feel fear because my heart races. Thus, part of an emotion is the representation of the bodily response which causes the emotion. This does not only include bodily processes that we can easily become aware of, such as heart racing or sweating, but also processes that we are largely unconscious of, such as hormone secretions and neurotransmitter discharges. These bodily changes are registered by the interoceptive system, which is therefore integral to emotion production.

However, emotions are more than just representations of bodily processes. In order to be able to tell emotions apart from one another, and from non-emotions (e.g. feeling cold), and to account for the intentionality of emotions, something more is needed.[1] This is why this theory also posits that representations of external objects are also part of the emotion. The reason why my fear of the snake is about the snake is thus because the intentional object is a representation of it. The difference between being scared of a snake and being scared of a lion can be explained with reference to the intentional object: in the first case, the content of the intentional object is a representation of a snake, and in the second case it is a representation of a lion. Integrating this representation of an external object with the representation of the bodily state enables us to explain both why the emotion is about a certain thing, and why it has a particular phenomenology. The reason why I feel scared of the snake is because of my bodily response being represented in a certain way, and the reason why the fear is about the snake is because of the representation of the external object.

How about moods? An attractive account of moods would be one which links it closely to emotions whilst explaining the phenomenal differences. A key observation about moods is that they differ slightly in phenomenal character from emotions; that they do not seem to be about anything in particular, whereas emotions do. Extending the Impure Somatic Theory to capture moods, my suggestion is that a mood is entirely constituted by the type of bodily representations that partly constitute an emotion. The only difference between a mood and an emotion is that a mood does not involve any representation of an external object. Thus, moods also consist of a representation of a bodily state, although the bodily processes that produce moods may be different from the bodily processes that produce emotions. This account thus explains why moods and emotions are very similar in phenomenology, yet different. What does this mean for our ability to forecast moods and emotions?

The cognitive architecture we are left with is one where a single affective system produces both moods and emotions, albeit slightly differently. In order to affectively forecast moods and emotions, then, the affective system, whose primary function is to produce moods and emotions, needs to be used in a secondary way. That is, when we imagine a mood or an emotion, we reuse the affective system to produce imaginary moods and emotions, using the same processes which are implemented in the same mechanisms (Goldman, 2006; Hurley, 2008). However, whereas actual emotions and moods are caused by a stimulus out in the world, the imaginary mood or emotion is caused by the agent’s intentionally imagining it. When I imagine how I would feel if I had an ice cream, I utilise what the emotion would be about in order to forecast it experientially. Imagining the intentional object – the representation of the ice cream – produces a sensory imagining of excitement. This is why we are most often accurate when imagining future emotions (Levine et al., 2013). Moods, on the other hand, have no intentional object, so we cannot use it to initiate the imagining.[2] A mood is not about anything, so we cannot simulate what it would feel like to be in a bad mood by using the intentional object of the mood, since there is none. In fact, any time we do initiate a forecast by imagining an intentional object, we automatically imagine an emotion. This reliable technique of forecasting is thus not available when trying to forecast moods. Since we can, however, accurately forecast emotions relatively easily, my suggestion is that instead of trying to imagine our future moods, we often resort to only imagining our future emotions, because this is a cognitively easier task. This conclusion fits with data from affective forecasting studies, which show that we imagine future moods to be more intense than they actually are. Since emotions are normally more intense than moods, and I have presented reasons to think it is cognitively difficult to forecast moods, it looks like the subjects in the studies have imagined their future emotions rather than their future moods (Kaplan et al., 2016).

If I am right, psychologists working on affective forecasting have wrongly assumed that we imaginatively forecast both moods and emotions. This is a problem, since many studies claim that we are bad at affective forecasting. The real reason why we appear bad might instead be because a purportedly forecasted mood (actually an emotion) has been compared to a real reported mood, leading researchers to conclude that we imagine our moods to be more intense than they actually turn out to be. I believe that what I have said in this post regarding the cognitive architecture of moods and emotions can inform the research on affective forecasting – and it might turn out that we are better than we think we are!


[1] See Cannon (1927) for this objection

[2] Note that I am not claiming that it is impossible to accurately forecast moods. It is possible to use other techniques to forecast moods. I could for example know about myself that I am normally in a bad mood after having been in three long meetings without any lunch, and therefore imagine that this will also be the case next time. But this is closer to inductive reasoning than the cases I am discussing.


References:

Barlassina, L. and Newen, A. (2014) ‘The Role of Bodily Perception in Emotion: In Defense of an Impure Somatic Theory’, Philosophy and Phenomenological Research, 89(3), pp. 637–678.

Cannon, W. B. (1927) ‘The James-Lange Theory of Emotions: A Critical Examination and an Alternative Theory’, The American Journal of Psychology, 39(1), pp. 106–124.

Gilbert, D. T. et al. (1998) ‘Immune Neglect: A Source of Durability Bias in Affective Forecasting’, Journal of Personality and Social Psychology, 75(3), pp. 617–638.

Goldman, A. (2006) Simulating Minds: The Philosophy, Psychology, and Neuroscience of Mindreading. Oxford University Press: USA.

Hurley, S. (2008) ‘Understanding Simulation’, Philosophy and Phenomenological Research, LXXVII(3), pp. 755–774.

James, W. (1884) ‘What is an emotion?’, Mind, 9, pp. 188–205.

Kaplan, R. L. et al. (2016) ‘Forgetting feelings: Opposite biases in reports of the intensity of past emotion and mood’, Emotion, 16(3), pp. 309–319.

Levine, L. J. et al. (2012) ‘Accuracy and artifact: Reexamining the intensity bias in affective forecasting’, Journal of Personality and Social Psychology, 103(4), pp. 584–605.

Levine, L. J. et al. (2013) ‘Like schrödinger’s cat, the impact bias is both dead and alive: Reply to Wilson and Gilbert (2013)’, Journal of Personality and Social Psychology, 105(5), pp. 749–756.

Paul, L. A. (2004) Transformative Experience. New York: Oxford University Press.

Prinz, J. (2004) Gut Reactions: A Perceptual Theory of the Emotions. Oxford University Press.

Wilson, T. D. et al. (2000) ‘Focalism: A source of durability bias in affective forecasting’, Journal of Personality and Social Psychology, 78(5), pp. 821–836.

Wilson, T. D. and Gilbert, D. T. (2000) ‘Affective Forecasting’, Advances in Experimental Social Psychology, 35, pp. 345–411.