Mixing Memory

According to many theories of embodied cognition (particularly type 5), perception is designed to facilitate bodily action, and therefore perception and movement are deeply connected. Much of the evidence for this position comes from research on the relationship between attitudes and movements. For example, Cacioppo, Priester, and Berntson1 showed that if people were presented with objects while they moved their arm in a way that mimicked pulling something towards them (an approach movement), they liked them more than if those same objects were presented while they moved their arm as though they were pushing something away (an avoidance movement).

In another study, Chen and Bargh2 had participants sit in front of a cathode ray tube while grasping a lever that could be moved forward and backward. The participants saw positive and negative words on the cathode ray tube’s screen, and were either instructed to move the joystick forward (away from their bodies; an avoidance movement) for negative words and backwards (toward their bodies; an approach movement) for positive words (this was the congruent condition, because the movements were consistent with their attitudes towards the words), or they were told to move the joystick forward for positive words and backwards for negative words (the incongruent condition). In two experiments, Chen and Bargh found that reaction times were faster in the congruent condition than in the incongruent condition (see the graph, from their Figure 1, p. 219).

i-ffb031e99f97949f980776cd996b174d-chenbarghfig1.JPG

Combined, the Cacioppo et al. and Chen and Bargh studies provide evidence for connections between perception and movement in both directions. The Cacioppo et al. study shows that perception (in this case, the perception of the value of an object) is influenced by movements (approach or avoidance), while the Chen and Bargh study shows that evaluative perception can influence movements (movements inconsistent with evaluations were slower). Embodied cognition theorists have thus argued that these studies and others like them provide evidence for one of the most central embodied cognition positions.

For various reasons, Markman and Brendl3 were unhappy with the perception-movement connection explanation for these results, so they conducted a series of studies similar to the Chen and Bargh experiments, but with a twist. Like Chen and Bargh, they presented participants with positive and negative words and asked them to move a joystick forward or backwards in consistent or inconsistent motions. The twist was that in addition to presenting the positive and negative words on the screen, the participant’s name was also on the screen. The positive and negative words were either presented in front of their name or behind it. Here’s what they saw, from Markman and Brendl’s Figure 1 (p. 8):

i-336f4669bbb1bd23218b9b904c771c7f-markmanbrendlfig1.JPG

Instead of being instructed to move the lever toward or away from themselves, they were instructed to move the lever toward or away from their names. Thus, in the consistent condition, if a positive word was presented behind their name, they would pull the lever back (forward for a negative word), while if a positive word was presented in front of their name, they would move the lever forward (backwards for negative words). The movements were reversed in the inconsistent condition. This means that in the consistent condition, the movements for words presented behind their name were identical to those in Chen and Bargh’s consistent condition (approach toward the body; avoidance away from it), but if the words were presented in front of their names, the consistent movements were the reverse of Chen and Bargh’s (approach away from the body; avoidance towards it).

In this experiment, if the embodied cognition view that perception and action are coupled, and embodied, is correct, then for words presented behind participants’ names, the results should be identical to Chen and Bargh’s, while there should be no difference between the consistent and inconsistent conditions, or a difference in the opposite direction from Chen and Bargh’s results, when the words are presented in front of participants’ names (because the movements are actually inconsistent in the consistent condition). Markman and Brendl did find that when the words were presented behind their names, so that the consistent motions involved pulling the lever towards their bodies for positive words and pushing the lever away from their bodies for negative words, reaction times were faster in the consistent and inconsistent condition. Inconsistent with the predictions of the embodied cognition theories, however, they found that when the words were presented in front of participants’ names, the consistent condition, which involved pushing the lever forward (toward their names) for positive words and pulling it back (away from their names) for negative words, produced faster reaction times than the inconsistent condition.

Markman and Brendl argue that this result shows that the connection between perception and action is dependent on where we reperesent the self in space. In the Chen and Bargh study, the self was represented in the same position as the body, while in the Markman and Brendl study the self was represented in the position of the name on the screen. Thus, approach and avoidance motivations resulted in movements that were faster to move positive words towards their name, and negative words away from their name, rather than towards or away from their bodies. According to Markman and Brendl, theories of embodied cognition can’t account for this representation of the self outside of the body without recourse to “higher-order symbolic processiong” (i.e., something other than traditional embodied representations and processes). They conclude:

The present results suggest that perceptual and motor representations alone may not be sufficient to account for cognitive processing, because phenomena that at face value seem prime examples of lower-order perceptual and motor processing may nonetheless involve higher-order symbolic processing. At a minimum, embodied accounts must specify how symbolic information about the self is tied to perceptual and motor representations.


1Cacioppo, J.T., Priester, J.R., & Bernston, G.G. (1993). Rudimentary determination of attitudes: II. Arm flexion and extension have differential effects on attitudes. Journal of Personality and Social Psychology, 65, 5-17.
2Chen, S., & Bargh, J.A. (1999). Consequences of automatic evaluation: Immediate behavior predispositions to approach or avoid the stimulus. Personality and Social Psychology Bulletin, 25, 215-224.
3Markman, A.B., & Brendl, C.M. (2005). Constraining theories of embodied cognition. Psychological Science, 16(1), 6-10.

Comments

  1. #1 Michael Anes
    November 28, 2006

    Hi Chris,
    Did the original Chen and Bargh study change response hands? I ask because approach and avoidance cognitions and behaviors are suggested to be hemispherically specialized by Davidson. If indeed the RH is specialized for avoidance-related behaviors, then left hand congruent away motions ought to be faster than right hand congruent away motions (and there are lots of other interesting comparisons)- are you aware of a combination of the hemispheric specialization literature and embodied cognition literature?

    Still chewing on the Markman study,
    M

  2. #2 Chris
    November 28, 2006

    Michael, no, I wasn’t aware of hemispheric specialization in the embodied cognition literature. Both Chen and Bargh and Markman and Brendl had participants use their dominant hand, which means that, in all likelihood, the vast majority used their right hands.

  3. #3 igor eduardo kupfer
    November 28, 2006

    Nice post. As a tangetnial observation, why are there so many bad graphs in the psych literature? Figure 1 above contains only four values — a horrible waste of ink. These numbers could be presented just as profitably within the text itself.

  4. #4 Chris
    November 28, 2006

    Well, I prefer graphic presentations to text, but that is a bad graph. I mean, no error bars. What the hell can you tell from a graph with no error bars?

    Also, notice that the scale is different on the two graphs. Ugh… social psychologists.

  5. #5 Michael Anes
    November 28, 2006

    This reminds me of the previous post here in which a statistics discussion is suggested, as well as the Frontal Cortex discussion of statistics as an educational requirement!

    Bad Graph. All persons should understand and justiably adore the Standard Error of the Mean. The SEM and the Central Limit Theorem should rock everyone’s world.

  6. #6 Chris
    November 28, 2006

    Definitely. I was thinking that I would take a lot of points off if a student handed me that graph.

    And a statistics primer definitely looks necessary.

  7. #7 igor eduardo kupfer
    November 28, 2006

    Oh,oh — if you’re going to do stats, can you talk about graphs? And factor analysis? And how I’m supposed to cluster these Likert scale data I have in front of me that my education didn’t prepare me for because I didn’t get to the stuff about ordinal and categorical data but I’m supposed to have prepared by the end of the year?

  8. #8 Chris
    November 28, 2006

    Hah… I think factor analysis might be taking blog education a bit too far.

  9. #9 mcewen
    November 28, 2006

    ‘Markman and Brendl argue that this result shows that the connection between perception and action is dependent on where we reperesent the self in space.’ One of my sons has ‘vision therapy’ which addresses that issue. It is fascinating to observe the therapist trying to ‘get him grounded’ so that his perceptual awareness and acuity is increase. Gotta see it to believe it [As we ‘pretend’ Americans say]
    Best wishes
    http://whitterer-autism.blogspot.com

The site is currently under maintenance and will be back shortly. New comments have been disabled during this time, please check back soon.