More about the significance of mental perturbance
A key concept for understanding the human mind
Last week on Substack I mentioned that a 1996 paper of ours on mental perturbance (“emotion”) has recently been “selected for inclusion in the 2026 four-volume reference work Artificial Intelligence: Critical Concepts in Cognitive Science —a collection intended to map the intellectual development of AI as a field contributing to cognitive science.” Mental perturbance is an information-processing condition in which insistent concerns capture and disrupt executive control over time. Along with Sylwia Hyniewska and Monica Pudło, I published a sequel to this paper in 2020: Mental perturbance: An integrative design-oriented concept for understanding repetitive thought, emotions and related phenomena involving a loss of control of executive functions.
I consider mental perturbance to be one of the key concepts necessary to understand the human mind. The situation with understanding perturbance is like the proverbial apple that fell on Newton’s head. Everyone knows repetitive thought is a key feature of human mind, but few people recognize that these phenomena can be understood as manifestations of a common architectural condition—mental perturbance. The situation is similar with respect to understanding art: everyone knows that repetition is a key feature of music, poetry and song. But few people realize repetition is used by artists because it resembles a key feature of mind: perturbance. Repetition in art can sustain activation of concerns or expectations without resolution, thereby inducing a controlled, aesthetic form of perturbance-like attentional capture. To understand mental perturbance requires an integrative design-oriented approach. It is defined in integrative design-oriented terms. (I’m not suggesting that the concept of perturbance has the same rigoror or maturity as Newton’s laws of motion.)
Margaret Boden (accolades of her here) recognized the importance of mental perturbance in her commentary on our 1996 paper:
The theoretical work of Wright, Sloman, and Beaudoin is a significant contribution to our understanding of the nature and function of emotions, and potentially also to therapeutic method. Their message that emotions, as controlling and scheduling mechanisms, are essential to any complex intelligent system (that is: one with multiple and potentially conflicting motives, and situated in a changing and unfriendly world) is important. It encourages cognitive psychologists to regard emotions as a central aspect of intelligence, not to sideline (or even to forget) them, as is too often done. In addition, it shows—at least in general outline—how emotions might be understood in computational terms.
Critics of computational psychology too often assume that emotions and motivation must lie outside its scope. This is not so, although it is true that relatively little computational work has been done in this area. Interestingly, the very early days of AI saw a comparatively high proportion of models of multiple goal-seeking, affective processes, and even personality (see, for example, the papers in Tomkins and Messick 1963). But, inevitably, these were so crude as to be almost useless in illuminating the nature of emotion and motivation. As AI-workers and their psychologist colleagues found to their cost, it is difficult enough to model the pursuit of one goal, never mind the pursuit of several mutually conflicting goals. Accordingly, personality and emotion soon faded from the scene as preferred topics of computational research—but not before I too had argued that they could, in principle, be so addressed (Boden 1965; 1972).
I suggested—in the sketchiest terms—that the concepts of William McDougall’s personality theory could be interpreted in a computational fashion (Boden 1965; 1972). His concept of sentiment, for example, marks what Wright et al. call “goal-sets,” and his master sentiment of self-regard relates to the descriptive and normative aspects of the self-image to which they draw our attention. He even described emotions as (conscious) control mechanisms, enabling the subject to monitor the incipient excitement of distinct types of motivation, and (sometimes) to adjust behavior so as to facilitate or inhibit the otherwise-inevitable response. What the target-authors have done is to put subtle and sophisticated computational flesh onto these minimalist theoretical bones.
McDougall himself, of course, will be turning in his grave. He saw affect as partly conscious (as well as dispositional), and argued that consciousness—and purposive behavior too—cannot possibly [End Page 135] be given a mechanistic, or physicalist, explanation. Despite Wright et al.’s persuasive anti-reductionist remarks about the necessity for highly abstract (psychological) levels of explanation, he would not have accepted that such explanations could be computationally grounded. He even insisted that the neurophysiological ground of goal-seeking behavior involved a special sort of energy, or horme, intrinsically directed to specific ends. Presumably, no professional reader today would agree with McDougall on that point. But many may agree with him about consciousness. And many more will insist that consciousness is a crucial aspect—perhaps the crucial aspect—of emotion. Such critics will doubtless point out that our three authors explicitly “factor out,” or refuse to discuss, consciousness as such.
Wright et al.’s rhetorical reserve on this point is defensible on two counts. First, even though emotions do involve conscious sensations, the dispositional and control aspects of affect are at least as important. Indeed, it is evident from the quotations from individuals reacting to bereavement that people often describe the phenomenology of grief in terms of these sorts of factors, rather than in a “purely qualitative” fashion (whatever that might be). Even so, these dispositional aspects of emotion are often overlooked, or regarded as less significant than the conscious phenomenology. And second, theoretical psychologists in general sideline consciousness: either they ignore it entirely, or they take it for granted philosophically and ask about the conditions under which it appears or disappears. They are well advised to do so, since we do not yet understand this concept (more accurately: this mixed bag of related concepts) at the philosophical level. Even a journal whose title has “Philosophy” as its leading word need not insist that every time consciousness is mentioned it must be philosophically discussed.
The specific example of grief, as the authors show, is a richly structured phenomenon, broadly distributed within the mind as a whole. As they also point out, it is essentially connected with attachment and personal love, from whose psychological complexity the complexity of grief is derived. Wright et al. are by no means the first to point out the complexities of love. Quite apart from the many intuitive presentations in literature, some philosophers who have analyzed the concept of personal love have shown that our everyday understanding of this concept assumes the presence of the sorts of dispositional factors highlighted in this paper (Fisher 1990). Similarly, these authors are not the first to describe the phenomenology and behavioral expressions of grief.
What they have done is to show–in relatively clear, and potentially testable terms–how this sort of psychological phenomenon is possible, and how the cognitive, affective, and motivational aspects of the mind are intimately combined in its very structure. That is, they have shown what sort of information-processing system a mind must be, in order that it may evince emotional phenomena of this type. In doing so, they have laid the groundwork for a systematic approach to therapy. And they have whetted our appetites–or at least, mine–for the implementational studies that are already being commenced in their research program.
Boden’s analysis highlights something that remains underdeveloped even today: the need to understand emotion as an architectural phenomenon involving executive function being captured by insistent motivators; that means the mind contains mechanisms to generate and activate motives (we call them “motive generators”.) The concept of mental perturbance extends this line of thought by explaining how such systems can enter sustained, self-maintaining states of attentional capture.
I mention this in the hope that people — in the cognitive/affective science and elsewhere — realize there’s a reason why apples fall from trees, and they experience repetitive thought and related phenomena. As it stands, most emotion theories do not deal with such attentional aspects of emotion. What is missing is not recognition of emotion, motivation, or even attention in isolation, but an account of how insistent concerns can dynamically capture and disrupt executive control over time. What is missing across these approaches is an account of how insistent concerns dynamically capture and disrupt executive control over time. For instance, Klaus Scherer’s Componential Process Model of Emotions does not include perturbance. I asked him at a conference whether he felt attentional aspects of emotion were important and he seemed to answer in the negative. I don’t mean to single him out, it’s just an example.
Herbert Simon, in a seminal paper on emotion, and Keith Oatley got into the ballpark of perturbance. However, they did not explicitly address emotion in integrative design-oriented terms, nor capture the dispositional properties of perturbance. Simon famously departed from his colleague, Alan Newell, and rejected the concept of information-processing architectures. Aaron Sloman, who with Monica Croucher in 1981 first developed the concept of perturbance (initially calling it emotion), was among the first to explicitly develop and emphasize the notion of information-processing architectures in this context.
Panksepp also got close with his distinction between primary, secondary, and tertiary processes, which implicitly points to an architectural layering of affective and cognitive systems. In describing tertiary process he alludes to the role of attention. However, his framework does not articulate how higher-level processes can give rise to sustained, control-disruptive states such as perturbance, nor does it develop an explicit information-processing architectural account of these dynamics.
What is missing is not recognition of emotion, motivation, or even attention in isolation, but an account of how insistent concerns can dynamically capture and disrupt executive control over time.
In sum, perturbance is not an optional feature of mind, but an emergent consequence of architectures that support multiple competing motives and flexible control. It is the architectural key to understanding how concerns become insistent, capture attention, and shape the flow of human experience over time.