No entity without identity: individuating attitudes in social psychology
Social psychologists like to write about attitudes. In fact, following Allport (1935), many of them have happily commented that the attitude is the most central and indispensable construct in social psychology (e.g., Petty, Wegener, Fabrigar, 1997). Here is a standard definition of an attitude: an attitude is
a psychological tendency that is expressed by evaluating a particular entity with some degree of favor or disfavor. (Eagly & Chaiken, 2007, p. 598)
A somewhat more specific view has it that attitudes are
associations between a given object and a given summary evaluation of the object — associations that can vary in strength and, hence, in their accessibility from memory. (Fazio, 2007, p. 608)
Attitudes are also supposed to be important for predicting behavior, though the attitude–behavior link is the subject of a great deal of controversy, which I can’t fully treat here. An extreme, design-oriented view is expressed by a B.F. Skinner-channeling B.J. Fogg:
While Fogg isn’t representative of mainstream, contemporary social psychology, similarly skeptical thoughts are expressed by investigators like Schwartz (2007). On the other hand, one common view of the attitude–behavior link is that it is quite strong (Kraus, 1997), but that (a) many research methods fail to measure attitudes and behaviors with regard to the same entities (Ajzen & Fishbein, 1977) and (b) this link is an important empirical subject, not built into the attitude construct by definition (Fazio, 2007; Zanna & Rempel, 1988).
I’ll set aside for now just how useful attitudes are for predicting behavior. But what should we make of this construct? That is, should we keep it around? Do we expect something like social psychology’s attitudes to be part of a mature science of human behavior?
Maybe I’m a sucker for a good slogan, but when I read psychologists’ on attitudes, I think of Quine’s slogan: no entity without identity. That is, we shouldn’t posit objects that don’t have identity conditions — the conditions under which we say that X and Y are the same object.
This slogan, followed strictly in everyday life, can get tricky: a restaurant changes owners and name — is it the same restaurant? But it is pretty compelling when it comes to the entities we use in science. Of course, philosophers have debated this slogan — and many particular proposed cases of posited entities lacking identity conditions (e.g., entities in quantum physics) — so I’ll leave it that lacking identity conditions might vary in how much trouble it causes for a theory that uses such entities.
What I do want to comment on is how strikingly social psychology’s attitudes lack good identity conditions — and thus have no good way of being individuated. While we might think this doesn’t cause much trouble in this case (as I just noted), I actually think it creates a whole family of pseudo-problems that psychologists spend their time on and build theories around.
First, evidence that there is trouble in individuating attitudes: As is clear from the definition of an attitude provided above, attitudes are supposed to be individuated by their object:
This evaluative responding is directed to some entity or thing that is its object—that is, we may evaluate a person (George W. Bush), a city (Chicago), an ideology (conservatism), and a myriad of other entities. In the language of social psychology, an entity that is evaluated is known as an attitude object. Anything that is discriminable or held in mind, sometimes below the level of conscious awareness, can be evaluated and therefore can function as an attitude object. Attitude objects may be abstract (e.g., liberalism, religious fiindamentalism) or concrete (e.g., the White House, my green raincoat) as well as individual (e.g., Condoleezza Rice, my sister-in-law) or collective (e.g., undocumented workers, European nations). (Eagly & Chaiken, 2007, p. 584)
So, for example, I can have an attitude towards Obama. This attitude can then have internal structure, such that there are multiple evaluations involved (e.g., implicit and explicit). This seems pretty straightforward: it is at least somewhat clear when some cognitive structures share the Obama as object.1
But trouble is not far around the corner. Much discussion of attitudes involves attitudes objects that are abstract objects — like sets or classes of objects– embedded in a whole set of relationships. For example, I might have attitudes towards snakes, Blacks, or strawberry ice cream. And there isn’t any obvious way that the canonical class by which attitudes are to be individuated gets picked out. A person has evaluative responses to strawberry ice cream, Ben & Jerry’s brand ice cream, ice cream in general, the larger class of such foods (including frozen yogurt, gelato, “soft serve”), foods that cool one down when eaten, etc.
This doesn’t just work with ice cream. (Obama instantiates many properties and is a member of many relevant classes.)
At this point, you might be thinking, how does all this matter? Nothing hinges on whether X and Y are one attitude or two…2
The particular trouble on my mind is that social psychologists have actually introduced distinctions that make this individuation important. For example, Eagly & Chaiken (2007) make much of their distinction between intra-attitudinal and inter-attitudinal structure. They list different kinds of features each can have and use this distinction to tell different stories about attitude formation and maintenance. I’m not ready to give a full review of these kinds of cases in the literature, but I think this is a pretty compelling example of where it seems critical to have a good way of individuating attitudes if this theory is to work.
Maybe the deck was stacked against attitudes by my prior beliefs, but I’m not sure I see why they are a useful level of analysis distinct from associations embedded in networks or other, more general, knowledge structures.
What should we use in our science of human behavior instead?
I’m surprised to find myself recommending this, but what philosophers call propositional attitudes — attitudes towards propositions, which are something like what sentences/utterances express — seem pretty appealing. Of course, there has been a great deal of trouble individuating them (in fact, they are one of the kinds of entities Quine was so concerned about). But their individuation troubles aren’t quite so terrible as social psychology’s attitudes: a propositional attitude can involve multiple objects without trouble, and it is the propositional attitudes themselves that can then specify the relationships of these entities to other entities.
I’m far from sure that current theories of propositional attitudes are ready to be dropped in, unmodified, to work in empirical social psychology — Daniel Dennett has even warned philosophers to be wary of promoting propositional attitudes for use in cognitive science, since theory about them is in such a mess. But I do think we have reason to worry about the state of the attitude construct in theorizing by social psychologists.
References
Ajzen, I., & Fishbein, M. (1977). Attitude-Behavior Relations: A Theoretical Analysis and Review of Empirical Research. Psychological Bulletin, 84(5), 8–918.
Allport, G. W. (1935). Attitudes. In C. Murchison (Ed.), Handbook of Social Psychology (Vol. 2, pp. 798–844). Worcester, MA: Clark University Press.
Eagly, A. H., & Chaiken, S. (2007). The Advantages of an Inclusive Definition of Attitude. Social Cognition, 25(5), 582-602.
Fazio, R. H. (2007). Attitudes as object-evaluation associations of varying strength. Social Cognition, 25(5), 603-637.
Fodor, J. A. (1980). Methodological solipsism considered as a research strategy in cognitive psychology. Behavioral and Brain Sciences, 3(1), 63–73.
Kraus, S. J. (1995). Attitudes and the Prediction of Behavior: A Meta-Analysis of the Empirical Literature. Pers Soc Psychol Bull, 21(1), 58-75. doi: 10.1177/0146167295211007.
Petty, R. E., Wegener, D. T., & Fabrigar, L. R. (1997). Attitudes and Attitude Change. Annual Review of Psychology, 48(1), 609-647.
Quine, W.V.O. (1969). Speaking of Objects. Ontological Relativity and Other Essays. New York: Columbia University Press.
Schwarz, N. (2007). Attitude Construction: Evaluation in Context. Social Cognition, 25(5), 638-656.
Zanna, M. P., & Rempel, J. K. (1988). Attitudes: A new look at an old concept. The Social Psychology of Knowledge, 315–334.
- There is still plenty of room for trouble, but this will be common to many representational constructs. For example, there are the familiar problems of what attitudes Louis has towards Superman. Superman is Clark Kent, but it would be odd if this external fact (which Louis doesn’t know) should determine the structure of Louis’ mind. See Fodor (1980). [↩]
- You would likely be in good company, I’m guessing this is a thought that was running through the heads of many of the smart folks in the seminar, “Attitudes and Persuasion”, in which I rambled on about this issue two weeks ago. [↩]
Unconscious processing, self-knowledge, and explanation
This post revisits some thoughts I’ve shared an earlier version of here. In articles over the past few years, John Bargh and his colleagues claim that cognitive psychology has operated with a narrow definition of unconscious processing that has led investigators to describe it as “dumb” and “limited”. Bargh prefers a definition of unconscious processing more popular in social psychology – a definition that allows him to claim a much broader, more pervasive, and “smarter” role for unconscious processing in our everyday lives. In particular, I summarize the two definitions used in Bargh’s argument (Bargh & Morsella 2008, p. 1) as the following:
Unconscious processingcog is the processing of stimuli of which one is unaware.
Unconscious processingsoc is processing of which one is unaware, whether or not one is aware of the stimuli.
A helpful characterization of unconscious processingsoc is the question: “To what extent are people aware of and able to report on the true causes of their behavior?” (Nisbett & Wilson 1977). We can read this project as addressing first-person authority about causal trees that link external events to observable behavior.
What does it mean for the processing of a stimulus to be below conscious awareness? In particular, we can wonder, what is that one is aware of when one is aware of a mental process of one’s own? While determining whether unconscious processingcog is going on requires specifying a stimulus to which the question is relative, unconscious processingsoc requires specifying a process to which the question is relative. There may well be troubles with specifying the stimulus, but there seem to be bigger questions about specifying the process.
There are many interesting and complex ways to identify a process for consideration or study. Perhaps the simplest kind of variation to consider is just differences of detail. First, consider the difference between knowing some general law about mental processing and knowing that one has in fact engaging in processing meeting the conditions of application for the law.
Second, consider the difference between knowing that one is processing some stimulus and that a various long list of things have a causal role (cf. the generic observation that causal chains are hard to come by, but causal trees are all around us) and knowing the specific causal role each has and the truth of various counterfactuals for situations in which those causes were absent.
Third, consider the difference between knowing that some kind of processing is going on that will accomplish an end (something like knowing the normative functional or teleological specification of the process, cf. Millikan 1990 on rule-following and biology) and the details of the implementation of that process in the brain (do you know the threshold for firing on that neuron?). We can observe that an extensionally identical process can always be considered under different descriptions; and any process that one is aware of can be decomposed into a description of extensionally identical sub-processes, of which one is unaware.
A bit trickier are variations in descriptions of processes that do not have law-like relationships between each other. For example, there are good arguments for why folk psychological descriptions of processes (e.g. I saw that A, so I believed that B, and, because I desired that C, I told him that D) are not reducible to descriptions of processes in physical or biological terms about the person.1
We are still left with the question: What does it mean to be unaware of the imminent consequences of processing a stimulus?
References
Anscombe, G. (1969). Intention. Oxford: Blackwell Publishers.
Bargh, J. A., & Morsella, E. (2008). The unconscious mind. Perspectives on Psychological Science, 3(1), 73-79.
Davidson, D. (1963). Actions, Reasons, and Causes. Journal of Philosophy, 60(23), 685-700.
Millikan, R. G. (1990). Truth Rules, Hoverflies, and the Kripke-Wittgenstein Paradox. Philosophical Review, 99(3), 323-53.
Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review, 84(3), 231-259.
Putnam, H. (1975). The Meaning of ‘Meaning’. In K. Gunderson (Ed.), Language, Mind and Knowledge. Minneapolis: University of Minnesota Press.
- There are likely more examples of this than commonly thought, but the one I am thinking of is the most famous: the weak supervenience of mental (intentional) states on physical states without there being psychophysical laws linking the two (Davidson 1963, Anscombe 1969, Putnam 1975). [↩]
Definitions of unconscious processing in cognitive and social psychology
John Bargh, Professor of Psychology at Yale, and his ACME (Automaticity in Cognition, Motivation, and Emotion) Lab are doing very exciting work. I had read some articles by Bargh some time ago (e.g. Bargh & McKenna 2004) and encountered his work in the context of debates about how objects can automatically activate attitudes that apply to them. But it hasn’t been until recently (following a discussion with James Breckenridge) that I’ve begun to really engage with the larger body of research Bargh and his collaborators have produced — and the interesting reflections and arguments found in the reviews of this and related work that he and his collaborators have written.
I expect I’ll be writing more about this work, but in this and some follow-up posts I want to just say a little bit about the general character of the research and, more specifically, how this work engages with and employs definitions of ‘unconscious’ and ‘unconscious processing‘.
Bargh & Morsella (2008, in press, page numbers are to this version) highlights how cognitive psychology and social psychology have operated with different definitions and different emphasis in investigating what they call “unconscious”. For cognitive psychology, “subliminal information processing – […] extracting meaning from stimuli of which one is not consciously aware” – has been paradigmatic of the unconscious (p. 1). That is, its study of unconscious processing is the study of the processing of stimuli of which one is unaware. On the other hand, for mainstream social psychology research, including work with priming, “the traditional focus has been on mental processes of which the individual is unaware, not on stimuli of which one is unaware” (Ibid.).
This is a striking difference that, as Bargh & Morsella illustrate, has consequences for how “dumb” or “smart” and “limited” or “pervasive” unconscious processing is. If unconscious processing is limited to processing of subliminal stimuli, then it doesn’t have much to go on. But the social psychology definition — the liberal, process-awareness definition — allows us to call a lot more things unconscious processing.
I recognize shortcomings with the cognitive psychology definition — the narrow, stimulus-awareness definition. And Bargh and Morsella’s statement of the process-awareness definition does enable them to say some striking things (e.g. about automatic activation of motivations).
But I also wonder whether this redefined term can bear much theoretical weight. Specifically, I have two concerns:
- this definition makes what is unconscious depend on each person’s knowledge of the causes of their actions — and this can get tricky in unintuitive and highly individual ways
- this definition seems to count on having good identity conditions for the kinds of objects to which ‘unconscious’ is supposed to apply (e.g. events, processes), but identity conditions (which are often hard to come by in general) are tricky for this domain in particular.
These are familiar problems in philosophy of mind, and they deserve consideration when designing theoretically useful definitions of unconscious processing. I aim to take up each of these issues in more detail in another post.
Bargh, J.A., & McKenna, K.Y.A. (2004). The Internet and social life. Annual review of psychology, 55, 573-590.