Tuesday, January 19, 2010

This aspirin is dictatorial, prosaic, and selfish

Fig. 2 (Schaefer & Rotte, 2010). Example of a questionnaire used to form a semantic differential for one particular brand. Subjects had to rate the brands according to their relationships to 18 pairs of contrary adjectives.

The nascent field of neuromarketing has grown tremendously in the last 5 years. Its goal is to use brain imaging techniques such as fMRI and EEG to gain hidden insights into consumer preferences. Commercial applications have far outstripped the peer-reviewed science necessary to validate their use. Google returns 322,000 hits for the term neuromarketing, whereas PubMed coughs up a sorry number - only 5 references! That doesn't stop large corporations from funneling millions of dollars into neuromarketing.

A new study published by Schaefer & Rotte, 2010 combined the techniques of semantic differentials (rating products along a 7 point scale between 2 bipolar adjectives, as shown in Fig. 2) with fMRI. First, participants viewed 18 pictures of common pharaceutical brands, such as aspirin. They were asked to rate these stimuli on a 5-point-scale regarding their personal attractiveness. After the imaging session was over, they completed the semantic differential questionnaire. These ratings were used to analyze the fMRI data.

Is this aspirin playful or serious?

Fig. 3 (Schaefer & Rotte, 2010). Semantic space built by the results of the semantic differentials. Brands and concepts are displayed on a 2D schema (factors ‘social competence’ and ‘potency’; factors were orthogonal). Colored circles depict the different brands. [NOTE: click on the figure for a larger view, and you can see the adjectives used in the present study are depicted in light blue.]

I knew next to nothing about semantic differentials before reading this paper, and that state of ignorance did not improve much when I finished reading. Some of the concepts were really murky and dependent on familiarity with the literature from 1957 and/or in German:
...Although semantic differentials have been investigated for decades, the correct description of these factors remains an issue. Thus, different researchers have been using very different labels for the underlying factors. This is particularly true for the factor ‘evaluation’ (Osgood et al., 1957), which also has been described as ‘mother’ [in a psychoanalytic view (Hofstätter, 1957)] and more recently as ‘social competence’ (Dziobek and Hülser, 2007).
The terms "evaluation", "mother", and "social competence" do not seem remotely equivalent. A book chapter by David Heise (1970) provided some clarity on the methodology:
(2) Ratings on bipolar adjective scales tend to be correlated, and three basic dimensions of response account for most of the co-variation in ratings. The three dimensions, which have been labeled Evaluation, Potency, and Activity (EPA), have been verified and replicated in an impressive variety of studies.

(3) Some adjective scales are almost pure measures of the EPA dimensions; for example, good-bad for Evaluation, powerful-powerless for Potency, and fast-slow for Activity. Using a few pure scales of this sort, one can obtain, with considerable economy, reliable measures of a person's overall response to something. Typically, a concept is rated on several pure scales associated with a single dimension, and the results are averaged to provide a single factor score for each dimension. Measurements of a concept on the EPA dimensions are referred to as the concept's profile.
So why is a brand of aspirin rated on the dimension of seclusive--sociable? Because the results here were incorporated into a larger semantic space established in an earlier study. Principal component analysis reduced the dimensionality of the data to two main factors that accounted for 87% of the variance: ‘evaluation’ and ‘potency’. The authors didn't like ‘evaluation’ for some unexplained reason and instead used the term ‘social competence’ for a series of inanimate pharmaceuticals. [NOTE: Guess I'm not a marketer...]

Moving on to the fMRI aspect of the experiment, it seems the authors wanted to relate the social competence of drug brands to neural activity in the medial prefrontal cortex (MPFC), which has been associated with self-referential processing and social cognition.

Fig. 4A (Schaefer & Rotte, 2010). Contrasts of brands loading high on the factor ‘social competence’ compared with brands loading high on the factor ‘potency’. This contrast showed significant activation in the MPFC and the SFG (superior frontal gyrus).

Oddly, the reverse contrast of high potency brands vs. high social competence brands did not produce any significant activity in the brain at all, which seems unexpected for a series of drugs. The high potency vs. low potency comparison revealed a reduction in SFG activity. To explain these results, the authors comitted the logical fallacy known as "reverse inference" by inferring the participants' mental state from the observed pattern of brain activity.
fMRI enable us to link the factors driven out of the behavioral data with the activation of certain cortical areas. Since the functional meaning of those areas (the MPFC and the SFG) are known from previous studies, the results can tell us what the extracted factors are about. More in detail, the results suggest to mark the factor originally described as ‘evaluation’ now as being mainly characterized by social perceptions. Thus, the description ‘social competence’ for this factor seems to be much more appropriate. Hence, the fMRI results provide important improvements for the factorial model of semantic space, which would not have been possible by looking on the behavioral data alone.
As for the SFG, they'd like to relate greater activity there to enhanced cognitive effort and working memory, but acknowledged this caveat:
However, since the SFG has been related not only to working memory but also to a variety of different cognitive functions, these explanations remain speculative.
Overall, I'm not sure how Bayer can target a new marketing campaign based on these results. The social competence of doctors recommending the aspirin, as in the classic commercial below? Comments, anyone?


Heise DR. (1970). The Semantic Differential and Attitude Research. Chapter 14 in Attitude Measurement. Edited by Gene F. Summers. Chicago: Rand McNally, pp. 235-253.

Schaefer, M., & Rotte, M. (2010). Combining a semantic differential with fMRI to investigate brands as cultural symbols. Social Cognitive and Affective Neuroscience DOI: 10.1093/scan/nsp055

Subscribe to Post Comments [Atom]


At January 21, 2010 10:17 AM, Anonymous SpGNo said...

On the other hand, there is no motivation for advocates of neuromarketing to publish in peer reviewed journals. If there are effective neuromarketing strategies, it's much better for the people who develop them to not publish them in peer reviewed journals.

At January 23, 2010 7:14 PM, Anonymous Anonymous said...

As usual, thanks for posting NC. Am I the only one to have noticed that SCAN publishes quite a bit of crapola these days? I'm not sure if they don't have good reviewers or what, but it's disturbing. Ultimately, this kind of stuff is what will cause the end of the social neuroscience (and whatnot) fad. About time.

At January 24, 2010 1:47 AM, Blogger Neuroskeptic said...

"Contrasts of brands loading high on the factor ‘social competence’ compared with brands loading high on the factor ‘potency’."

I'd have compared high "social competence" to low "social competence"... but I'm not a neuromarketer.

At January 29, 2010 11:18 AM, Blogger The Neurocritic said...

SpGNo - A good point, but how do they make an unbiased determination of "effectiveness"?

Anonymous - Well, here's a new fad that has just been published in NeuroImage: Live face-to-face interaction during fMRI: a new tool for social cognitive neuroscience.

Neuroskeptic - That would be too logical, because here's what they actually found:

"For further analysis, we chose brands that loaded high on the factor ‘potency’ (positively and negatively) and brands that loaded high on the factor ‘social competence’. However, almost no brands revealed negative loadings on the factor ‘social competence’ (see Figure 3). Thus, for this dimension we only used brands that loaded positively on this factor."


Post a Comment

<< Home

eXTReMe Tracker