We are in sympathy with the constructionist view of emotions put forward in the target article, “The brain basis of emotion: A meta-analytic review.” The results of Lindquist et al.'s meta-analysis are consistent with other comparative imaging studies, and with what is known about the inherently plastic exploration–stabilization processes that occur in the nervous system during development and learning. More generally, the results are in line with the growing evidence that commonalities at high levels of organization are underlain by variable networks that have only family resemblance with each other: The components of the networks provide facilitating rather than necessary or sufficient conditions for macro-level properties. We want to suggest two ways in which the authors' interpretation of the data can be supplemented. First, we suggest adding to their model a factor that helps to explain what Barrett (Reference Barrett2006b) has called the “emotion paradox” – the discrepancy between the perceived similarity that subjects and observers find among instances of “basic emotions” and the extreme variability among the neural and physiological correlates of such instances. Second, we expand on the role of language in the construction of human emotional categories.
We think that Lindquist et al.'s model does not provide a full solution to the emotional paradox. The analogy with color categories, which are also constructed categories, is not convincing because the biological affordances in this visual system – specific receptors for different wavelengths – do not have a counterpart in the model the authors describe. We believe that such a counterpart does exist, and that it is what James referred to as “instinct”: evolved behavioral dispositions, which, when the behavior they drive is first elicited, are the developmental scaffolds for constructing habitual motor behaviors and emotions (James Reference James1890/1950, Ch. 24). We regard Jamesian “instincts” as partially genetically assimilated behavioral dispositions that evolved in ways that make their typical behavioral manifestations require relatively little (initial) learning.
Examples in mammals are freezing, appetitive behavior, and so on. In normal animals such behaviors, as they are practiced, are constituted by specific coordinated motor reactions that have feedback relations with bound visceral sensations – with bodily experiences. In other words, initially, stereotypic behavior patterns inevitably have more or less stereotypic affective correlates – “first-time-feelings,” which are, for example, the feelings that arise as a mouse freezes for the first time, or as a hen first sits on an egg (see James Reference James1890/1950, Chs. 24 and 25). First-time feelings are not emotional categories, and, although similar, they are not identical among individuals of a species (because of the nature of early development). We think that they act as scaffolds for what, on the basis of ontogenetic experience and learning, later become a particular emotional category (e.g., freezing-fear). According to our view, an experienced, mature mouse has more than a general initial core affect; it has developmentally constructed emotional categories, although these categories are much narrower than those of humans, and are not adequately described in terms of human emotion-words (such as fear or love). They are better described in terms linked to the stereotypic responses, so a mouse might be said to have a freezing-emotion, or a maternal-care-emotion. These emotional categories are usually similar to but never identical to first-time feelings: The specific first-time feeling of freezing becomes enriched and fine-tuned during the ontogeny of each mouse as it interacts with predators (Ginsburg & Jablonka Reference Ginsburg and Jablonka2007a; Reference Ginsburg and Jablonka2007b; Reference Ginsburg and Jablonka2010a). In some cases the emotion associated with such scaffolds may be changed beyond recognition: for example, a male dog can be conditioned to react with an avoidance reaction to a bitch in heat (Gottlieb Reference Gottlieb1992), but in nonhuman animals this usually happens only under extreme or atypical conditions.
We therefore agree with Lindquist et al. that the emotions of nonhuman animals, like those of humans, are constructed during ontogeny from the stream of constantly evaluated homeostatic feelings (core-affect). These are what Lamarck called “inner feelings” (Lamarck Reference Lamarck1809/1914, Book III, pp. 333–34), and we believe they appeared very early in the evolution of animals (Ginsburg & Jablonka Reference Ginsburg and Jablonka2010a; Reference Ginsburg and Jablonka2010b). However, we think that as the feelings that are the consequence of the activities occurring as “instinctive” responses are generated, they form one of the inputs that construct emotional categories. These initial affective dynamic states form only a developmental scaffold, but the “structures” built on this scaffold usually have sufficient family-resemblance to be recognized as belonging to a particular emotional category. Taking them into account renders the “emotion paradox” less paradoxical.
Although inborn predispositions may help in explaining the canalization of non-symbolically mediated stereotypic expressions of emotions, we share Lindquist et al.'s view that the tokens covered by emotional categories constructed by language, which include a very wide spectrum of behaviors and subjective states, may have very little in common with each other except for their symbolic references. We would like to expand on this point from an evolutionary point of view. The evolution of language, which led to the ability to communicate about the not-here and the not-now, engendered profound alterations in the emotional profile of humans. This was the consequence of, first, the expanding breadth of an individual's experiences brought about by sharing the experiences of others through linguistic communication; second, the accompanying development of personal autobiographies, which increased the coherence and specificity of individual experiences; third, the formation of social values and new social emotions; and fourth, the control (often the inhibition) of the affective triggers of action, of affect-related drives through linguistic communication. Although the control of drives has ancient roots related to the evolution of episodic memory, tool use, and recall-for-planning, the human ability to communicate about the imagined necessitated an increase in this inhibitory capacity and in its regulation, and allowed the metaphorical transfer of symbols, creating new categories in all domains of human experience. All these led to emotional categories, which, as Lindquist et al. convincingly argue, are held together by emotion-words, and can be captured adequately only by symbols.
We are in sympathy with the constructionist view of emotions put forward in the target article, “The brain basis of emotion: A meta-analytic review.” The results of Lindquist et al.'s meta-analysis are consistent with other comparative imaging studies, and with what is known about the inherently plastic exploration–stabilization processes that occur in the nervous system during development and learning. More generally, the results are in line with the growing evidence that commonalities at high levels of organization are underlain by variable networks that have only family resemblance with each other: The components of the networks provide facilitating rather than necessary or sufficient conditions for macro-level properties. We want to suggest two ways in which the authors' interpretation of the data can be supplemented. First, we suggest adding to their model a factor that helps to explain what Barrett (Reference Barrett2006b) has called the “emotion paradox” – the discrepancy between the perceived similarity that subjects and observers find among instances of “basic emotions” and the extreme variability among the neural and physiological correlates of such instances. Second, we expand on the role of language in the construction of human emotional categories.
We think that Lindquist et al.'s model does not provide a full solution to the emotional paradox. The analogy with color categories, which are also constructed categories, is not convincing because the biological affordances in this visual system – specific receptors for different wavelengths – do not have a counterpart in the model the authors describe. We believe that such a counterpart does exist, and that it is what James referred to as “instinct”: evolved behavioral dispositions, which, when the behavior they drive is first elicited, are the developmental scaffolds for constructing habitual motor behaviors and emotions (James Reference James1890/1950, Ch. 24). We regard Jamesian “instincts” as partially genetically assimilated behavioral dispositions that evolved in ways that make their typical behavioral manifestations require relatively little (initial) learning.
Examples in mammals are freezing, appetitive behavior, and so on. In normal animals such behaviors, as they are practiced, are constituted by specific coordinated motor reactions that have feedback relations with bound visceral sensations – with bodily experiences. In other words, initially, stereotypic behavior patterns inevitably have more or less stereotypic affective correlates – “first-time-feelings,” which are, for example, the feelings that arise as a mouse freezes for the first time, or as a hen first sits on an egg (see James Reference James1890/1950, Chs. 24 and 25). First-time feelings are not emotional categories, and, although similar, they are not identical among individuals of a species (because of the nature of early development). We think that they act as scaffolds for what, on the basis of ontogenetic experience and learning, later become a particular emotional category (e.g., freezing-fear). According to our view, an experienced, mature mouse has more than a general initial core affect; it has developmentally constructed emotional categories, although these categories are much narrower than those of humans, and are not adequately described in terms of human emotion-words (such as fear or love). They are better described in terms linked to the stereotypic responses, so a mouse might be said to have a freezing-emotion, or a maternal-care-emotion. These emotional categories are usually similar to but never identical to first-time feelings: The specific first-time feeling of freezing becomes enriched and fine-tuned during the ontogeny of each mouse as it interacts with predators (Ginsburg & Jablonka Reference Ginsburg and Jablonka2007a; Reference Ginsburg and Jablonka2007b; Reference Ginsburg and Jablonka2010a). In some cases the emotion associated with such scaffolds may be changed beyond recognition: for example, a male dog can be conditioned to react with an avoidance reaction to a bitch in heat (Gottlieb Reference Gottlieb1992), but in nonhuman animals this usually happens only under extreme or atypical conditions.
We therefore agree with Lindquist et al. that the emotions of nonhuman animals, like those of humans, are constructed during ontogeny from the stream of constantly evaluated homeostatic feelings (core-affect). These are what Lamarck called “inner feelings” (Lamarck Reference Lamarck1809/1914, Book III, pp. 333–34), and we believe they appeared very early in the evolution of animals (Ginsburg & Jablonka Reference Ginsburg and Jablonka2010a; Reference Ginsburg and Jablonka2010b). However, we think that as the feelings that are the consequence of the activities occurring as “instinctive” responses are generated, they form one of the inputs that construct emotional categories. These initial affective dynamic states form only a developmental scaffold, but the “structures” built on this scaffold usually have sufficient family-resemblance to be recognized as belonging to a particular emotional category. Taking them into account renders the “emotion paradox” less paradoxical.
Although inborn predispositions may help in explaining the canalization of non-symbolically mediated stereotypic expressions of emotions, we share Lindquist et al.'s view that the tokens covered by emotional categories constructed by language, which include a very wide spectrum of behaviors and subjective states, may have very little in common with each other except for their symbolic references. We would like to expand on this point from an evolutionary point of view. The evolution of language, which led to the ability to communicate about the not-here and the not-now, engendered profound alterations in the emotional profile of humans. This was the consequence of, first, the expanding breadth of an individual's experiences brought about by sharing the experiences of others through linguistic communication; second, the accompanying development of personal autobiographies, which increased the coherence and specificity of individual experiences; third, the formation of social values and new social emotions; and fourth, the control (often the inhibition) of the affective triggers of action, of affect-related drives through linguistic communication. Although the control of drives has ancient roots related to the evolution of episodic memory, tool use, and recall-for-planning, the human ability to communicate about the imagined necessitated an increase in this inhibitory capacity and in its regulation, and allowed the metaphorical transfer of symbols, creating new categories in all domains of human experience. All these led to emotional categories, which, as Lindquist et al. convincingly argue, are held together by emotion-words, and can be captured adequately only by symbols.