In a 2010 paper, Spee Kosloff and colleagues report several studies involving political smears that they conducted during the month before the 2008 election. The smears were that Obama is a closet Muslim extremist and that McCain is senile.
One of the studies measured an effect that worked entirely below the level of consciousness. Participants were presented with strings of letters that were either words or nonwords (of English) and they pressed one of two buttons to indicate whether the string was or was not a word. Before they saw the letter string, two other things happened. (1) They saw the word “trial” (in the same place where the letter string would appear) for about three quarters of a second – amply long enough to see it and read it. (2) Between the “trial” and the letter string, there was a very brief exposure (28.5 thousandths, or about 1/35, of a second) of either “Obama” or “McCain”, also where the word “trial” had been and where the letter string would immediately appear. This exposure is too brief to read; most participants were unaware that there had been a word flashed between “trial” and the string to be classified as a word or nonword. (The roughly 15% who detected that there had been a word reported having had no idea what it was.)
Among the letter strings that were words, most had no relevance to political smears (e.g., “rectangle”, “lamp”), but a few were laden with such relevance. They were either Muslim-related terms, e.g., “Koran”, “mosque”, or senility-related terms, e.g., “dementia”, “Alzheimers”.
The measure in this study was the time it took from the onset of one of the laden words to the participant’s decision that it was a word. (Only correct decisions about word status were included in the data to be analyzed.)
The key results of this experiment are these. (a) Obama supporters decided that senility-related terms were words faster after “McCain” had been briefly flashed than after “Obama” had been briefly flashed. (b) Obama supporters were also faster than McCain supporters to decide that senility-related terms were words after “McCain” had been flashed. (c) and (d) are parallel results for McCain supporters and decisions about Muslim-related terms after the brief presentations of “Obama”.
In short, a presentation of a word too briefly to be consciously read can cause a measurable difference in the time it takes to decide that a politically laden word is a word. And this difference depends on the relation between the flashed word and one’s political views.
It may be tempting to downplay the significance of this result. It’s a special case, one might say. The task is artificial. The differences in reaction time are smaller than time differences that are relevant to real world action.
But I think that such a dismissive reaction would be unfortunate. The special case and the artificial experimental set up are necessary to get a clear observational result. But the conclusion that the evidence thus gained supports is that an unconscious stimulus can engage a cognitive process (i.e., one’s views about a candidate) and can do so entirely outside of consciousness. The lesson I am inclined to draw from this study is that we have some direct evidence that unconscious processes can have cognitive richness.
A second experiment tested the effect of making race salient (and consciously so) on a group of participants who indicated that they were undecided as to which candidate they supported.
The experimental task was, again, to decide whether a letter string was a word or a nonword. Among those strings that were words, most were neutral fillers, but a few were Muslim-related terms. The measure was the time from onset of the letter string to the pressing of the button indicating the word or nonword decision. The only decisions of interest are those that correctly classified Muslim-related words as words.
The key manipulation was that immediately before the decisions on letter strings began, participants filled out a questionnaire about themselves, which either did or did not include a question about race. This question provided six racial categories and asked participants to circle those “that are personally relevant to your identity”. No participants circled “African American”. Those who got the question are the “race salient” group, with the remainder being the non race salient group.
Participants saw a readable word, “trial”, followed by a too-brief-to-read, 1/35 of a second exposure of “McCain” or “Obama”, followed by the letter string to be classified as a word or nonword.
The most interesting, and somewhat disturbing, results of this experiment are these. (a) Undecided participants who were briefly exposed to “Obama” and who had answered the race question were faster to correctly classify Muslim-related words as words than similarly exposed undecided participants whose form did not include the race question. (b) This difference was not present when the briefly exposed word was “McCain”.
Since the smear that Obama is a closet Muslim extremist was often repeated prior to the 2008 election, it is presumed that all participants were aware of it. It appears from this study that this background awareness did not become activated if the matter of race had not been very recently made salient. But if it was made salient, then it was sufficiently activated to quicken the response on the word status decision task.
The race question itself was, evidently, consciously processed by those whose questionnaire included it. But the decision task was done rapidly and there is no question of the participants having consciously deliberated about the relation between race and the word status decision. So, even though the salience of race worked through a conscious input, the process by which it reduced decision time worked outside of conscious deliberation.
Once again, it might seem that the results of these two experiments show something about unconscious processing, but are unimportant for larger life because the differences in reaction times are far smaller than the time it takes us to think of and decide to execute any conscious, deliberate action (including, as always, speaking or indicating intent with a gesture). For example, the fact that a decision about a word took a fraction of a second less would not show that the decision was any different from what it would have been without the brief exposures or the inclusion of the race question. A third study, however, casts doubt on such a hopeful view.
In the third study, participants read articles of about 600 words, one that elaborated on the Obama smear and one that elaborated on the McCain smear. The articles were written by the experimenters and designed to be parallel in the types of support offered for Obama’s closet Muslim extremism or McCain’s senility, but they were produced in a format that made them look like copies of newspaper articles. After reading one or the other, participants were asked to rate their degree of endorsement of the thesis of the article they read.
Data were analyzed separately for those who had identified as Obama supporters, McCain supporters, or undecided. Care was taken to ensure that the participants knew that the experimenters would not be able to connect the responses to individual participants.
The key manipulation was that a questionnaire about the participants’ demographics, given immediately before the rest of the study, either included or did not include the race question (the same as in experiment 2) or a similar question about participants’ age group.
As expected, declared supporters of each candidate gave low endorsement to the smear of their candidate and higher endorsement to the smear of their candidate’s opponent. A key finding emerged from the results for the undecideds. Those who had received the race question gave higher ratings to the Obama smear than those who had not, and those who had received the age question gave higher ratings to the McCain smear than those who had not. The authors conclude that “It appears that undecided individuals can become motivated to accept smears of multiple candidates when situational factors render intergroup differences salient.” (p. 392)
This experiment is evidence for a sobering result. That one accepts a certain racial classification or a certain age classification as applying to oneself cannot be a reason for accepting or rejecting a smear of a candidate. The race and age of the candidates was well known to everyone. The manipulation consisting of including or not including the self-classification question regarding race or age did not supply new knowledge or a reason. Nonetheless it had an effect. It seems that this effect, therefore, worked through a process that did not engage conscious processing of the kind we would recognize as weighing reasons or evaluating evidence. Thus, this experiment provides evidence that even when inputs (reading and answering the race or age classification question) and outputs (making a mark on a scale indicating endorsement of an article’s thesis) are fully conscious, there can be processes that work outside of consciousness, yet produce effects on the conscious output.
[Kosloff, S., Greenberg, J., Schmader, T., Dechesne, M. and Weise, D. (2010) “Smearing the Opposition: Implicit and Explicit Stigmatization of the 2008 Presidential Candidates and the Current U. S. President”, Journal of Experimental Psychology: General 139(3): 383-398. This paper contains several results not stated here, and a fourth experiment that confirms and extends the results of experiment 3.]