Summary of &Quot;Brain Wave Data As an Advertising Diagnostic&Quot;

Michael L. Rothschild, University of Wisconsin
[ to cite ]:
Michael L. Rothschild (1983) ,"Summary of &Quot;Brain Wave Data As an Advertising Diagnostic&Quot;", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 184.

Advances in Consumer Research Volume 10, 1983      Page 184

SUMMARY OF "BRAIN WAVE DATA AS AN ADVERTISING DIAGNOSTIC"

Michael L. Rothschild, University of Wisconsin

In recent years the notion of low involvement learning has become increasingly popular in the advertising and marketing literature. In part this notion posits that consumers generally pay little attention to advertising (especially that which is broadcast) and, as a result, do not actively search for and process information. It is, therefore, questionable as to whether consumers can accurately respond to paper and pencil inquiries concerning what they have learned from such advertising.

Given the problems of memory search and verbalization one would desire a method which related to learning but minimized or avoided such tasks. Physiological measures (which elicit responses concurrent to exposure to the stimulus) are a set which potentially meet these criteria. Within this set, electroencephalic (brain wave) measures have been put forth as likely candidates.

Electrical activity continuously occurs in the of all people. The frequency and magnitude of this activity changes as a function of internal and external variables, can differ at different areas of the brain at the same time, can occur at different frequencies at the same time at the same brain location and can differ over time. Little is known as to how or why frequency and magnitude shift but information is increasing which shows that changes occur in predictable ways to various stimuli and tasks. There are also predictable differences across groups of people differentiated by variables such as age. sex and handedness.

EEG activity is a measure of activity taking place below a particular location on the skull, is typically measured at the surface of the scalp, amplified and recorded. Measured frequencies are typically in the range of 1 to 100 Hz (cycles per second) although generally one is concerned with frequencies under 30Hz. Amplitude is recorded within predetermined frequency ranges.

The most commonly observed frequency is "alpha" which varies between 8 and 13Hz with waves of 25 to 100 UV (microvolts) appearing mainly from mid and rear skull derivations. It has been often observed that the level of alpha rhythm is negatively related to many types of cognitive activity in that more cognitive responses to paper and pencil measures seem to occur when the level of alpha activity is suppressed. It is the nature of alpha that it is momentarily suppressed by the onset of a visual stimulus, and that recovery will begin immediately following its suppression. The rate of recovery is related to the content of the stimulus and interest in this stimulus. Auditory stimuli produce much less suppression of alpha, faster recovery and more rapid habitation.

Because of the well documented inverse relationship between alpha and cognitive activity/attention, alpha frequency is the most often studied and reported dependent variable when examining various types of cognitive activity, tasks and stimuli.

The EEG represents a continuous time graph of the electrical potential difference between a location of activity and a neutral reference point. Because of its continuous nature, a tremendous amount of response information is acquired in a very short time. This amount increases, of course, when observations are made at several locations simultaneously. A third dimension (in addition to time and location) which increases the amount of data concerns the several possible frequency bands from which one can sample.

To aid in controlling the volume of data, several steps can be taken. First the EEG signal can be passed through a set of filters which limit the data. For example, if alpha were being considered, 8 to 13Hz data would be kept and other frequencies filtered out.

Secondly, a Fourier transformation can be performed on the data. This converts activity within predetermined frequency range(s) to a measure of power. The power measure can be in voltage units or arbitrary units but in either case is easier to evaluate than the original time based data. Power is a measure of aggregate activitY in a location and/or frequency band.

Finally, the data which are now limited in frequency ranges and transformed to power units can he aggregated over time. In this way one can see units of power per frequency range per period of time. The minimum period of time must be great enough to allow the lowest frequency to be observed adequately.

The above and other issues were discussed as introduction to the reporting of a study of EEG data collected as 26 women watched a videotape of nine commercials imbedded in program content. The data show strong negative rank order correlations across the commercials when comparing total electrical activity in the alpha frequency band with recall and recognition. That is, a drop in alpha is related to greater correct responses in paper and pencil learning tests. A relationship between electrical activity and four adjective factors also exists; this relationship is in the correct direction but not significant. These findings are consistent with those in the psychophysiology literature based on less complex stimuli.

Data also show a relationship between sharp drops in alpha activity and visual scene changes. This relationship is also consistent with findings in the literature. By comparing these drops in alpha and the subsequent rate of alpha recovery with scenes in the commercials, one can observe which scenes are getting and/or holding attention better or more poorly than others. In this way one can use electrical activity as a diagnostic of. television commercials.

----------------------------------------