Perturbance: An Important Concept for Understanding “Racing Minds” and Other Forms of Repetitive Thinking

My colleagues and I recently published a paper that provides one of the essential concepts for understanding insomnia and a host of other constructive, banal, and problematic mental states. The concept is perturbance, which is a persistent tendency to consider affectively laden mental content — problems, issues, concerns, wishes, wants, desires, fears, yearnings, etc. Perturbance, however, is not simply defined in terms of folk psychology. It is meaningful and interesting because it involves a computational theory of mind.

Psychologists tend to believe that the main cause of insomnia is “racing minds”. Unfortunately, psychology lacks a deep, information-processing theory of “racing mind”. A helpful article on the subject is Watkins’ “Constructive and unconstructive repetitive thought”. (DOI: 10.1037/0033-2909.134.2.163). Watkins’ article tries to unify the literature on “rumination”, “worry”, “intrusive thought”, “perseverative thinking”, and to a lesser extent “obsession”. My colleagues and I agree it is important to unify these concepts. However, we believe the unification requires a deep information processing theory of mind, one that is explicit about the mechanisms involved.

First a bit of history. The concept of perturbance pre-dates the bulk of the growing psychological literature on “intrusive thought”. The concept comes from Prof. Aaron Sloman, a British pioneer in Artificial Intelligence research on emotion (and many other topics!). (Here’s a Wikipedia page about him. And his website, which I suspect is the most extensive website of original research of any AI researcher in the world.) I had the privilege of doing my Ph.D. with Sloman. I studied motivation in information processing terms. My Ph.D. thesis explained processes relevant to perturbance. One of my goals is to help psychologists understand that “repetitive thought” needs to be understood in terms of a designer-based theory of mind and of perturbance.

Here are some excerpt from our paper to whet your appetite:

In this paper, we argue that perturbance is a major feature of the human mind that has been ignored by psychologists but deserves considerable attention. This concept has the potential to unify several areas of study, including fundamental processes such as attention, emotion and emotion regulation, cognitive phenomena such as intrusive thought, and psychopathological conditions such as rumination, obsessive worrying and addictions. Like any theoretical concept, the concept of perturbance does not stand alone. It is meaningful, promising and useful because of the theoretical framework within which it is embedded: a) the CogAff architecture schema, and b) H-CogAff, a particular architecture based on CogAff which is aimed specifically at understanding humans [3].

Whereas Sloman made significant attempts to disseminate the design-based approach and H-CogAff to emotion and AI researchers, the impact on the psychology literature so far has been minimal, due to various factors some of which we will allude to here. Meanwhile, affective computing (AC), a discipline of computer science that focuses on emotion, including emotion modelling, is gaining momentum. However, AC currently tends to pursue narrow problems relevant to practical applications focusing on primary emotions (e.g., machine perception of primary emotions). In AC, there is almost no research on automatically detecting perturbance, let alone attempts to produce systems that can experience and monitor perturbance. This is the case despite the fact that Sloman’s work, including the concept of perturbance, was described in Rosalind Picard influential Affective Computing [65]. Sloman, who was one of the first AI researchers to systematically emphasize computational architectures, did foresee that AC would be a long road [66]. Still, AI’s highly visible progress, and its work on architectures, bode well for AC. We believe that history will prove Sloman’s theory of perturbance is a “sleeping beauty”. According to [8] these “beauties” tend to ‘awaken’ when they are discovered by a new community of researchers.

This paper is meant to promote consideration of H-CogAff by indicating its relevance to many phenomena and research communities, while focusing on one of its original concepts, perturbance. However, we only have space for a cursory overview of the theory itself. For more information about it, see [2-7,9-10] and other papers cited below.

[…]

Perturbance is of considerable adaptive significance because it is an affection of the human brain’s executive processes, which govern the agent.

We also believe a theory of perturbance can be used for positive psychology and self-help. For example, Beaudoin (2013) developed the cognitive shuffle a technique to combat insomnia which is meant to work partly by interfering with bedtime perturbance [52]. Focusing and flow are essential to cognitive productivity and hence to knowledge economies. Distraction is largely affective yet theories of attention —and knowledge translation on the subject e.g. [53-54] Levitin (2014), Gallagher (2006) — do not deal with motive processing and fail to invoke perturbance. Theories of learning, expertise and productive practice need to explain how humans can deliberately develop their mental architectures, e.g., creating new goal generators [47,55-56].

In short, previous research phenomena and problems can systematically be revisited from the designer stance as involving perturbance.

Our paper is somewhat dense and technical. However, it cites papers where interested readers can learn more. I intend to blog more about perturbance and this article on the CogZest website.

If you are interested in emotions and curious about how Artificial Intelligence research can illuminate our understanding of emotion, and provide guidance to psychology, and you happen to be in England in April of next month, check out the conference. Aaron Sloman will be giving a keynote lecture to the entire conference, and one to the Symposium on Computational Modelling of Emotion: Theory and Application. (Here is a link to the latter keynote by Aaron Sloman.) If you’re not in the area but interested, you can still obtain the peer-reviewed proceedings.

References

• Beaudoin, L. P., Hyniewska, S., & Hudlicka, E. (2017). Perturbance: Unifying Research on Emotion, Intrusive Mentation and Other Psychological Phenomena with AI. Paper to be presented at the Symposium on Computational Modelling of Emotion: Theory and Application at AISB-2017. Paper available from http://summit.sfu.ca/item/16776.
• Beaudoin, L.P. Goal Processing in Autonomous Agents
• Sloman, A. (2017). Architectures underlying cognition and affect in natural and artificial systems. Paper available from Aaron Sloman’s website.
• Watkins, E. R. (2008). Constructive and unconstructive repetitive thought. Psychological Bulletin, 134(2), 163–206.