Suppression-induced forgetting on a free-association test

Abstract

The repeated suppression of thoughts in response to cues for their expression leads to forgetting on a subsequent test of cued recall (Anderson & Green, 2001). We extended this effect by using homograph cues and presenting them for free association following suppression practice. Cue–target pairs were first learned under integrating imagery instructions; then in the think/no-think phase students practised suppressing thoughts connected to some homograph cues, with or without the assistance of thought substitutes that changed their meaning. Below-baseline forgetting on the subsequent free-association test was found in the production of suppressed targets. Following aided suppression this effect was also obtained in the production of other responses denoting the target-related meaning of the homograph cues. Discussion emphasises the ecological value of the test; rarely do people deliberately attempt recall of unwanted thoughts.

Keywords:

• Forgetting
• Implicit memory
• Suppression
• Inhibition
• TNT

Acknowledgments

We thank Allison Ford for assistance with both experiments.

Notes

¹Students were screened on the basis of their repressive orientation by modifying the system used by Myers and Derakshan (2004); we chose individuals who fell below the median on a test of trait anxiety and above the median on a test of social desirability. The use of this screening procedure was informed by Hertel and McDaniel's (2009) finding that repressors produced superior below-baseline forgetting.

²The data from one male student in the unaided suppress condition were set aside, due to self-reported depression at the end of the session. (See Hertel & Gerstle, 2003, for evidence that depressed students fail to show SIF.)

³Suppressed targets were recalled less well than baseline targets, F(1, 32)=22.54, MSE=170.15, p<.001, η_p ²=.41. Furthermore, SIF depended on the suppression method, F(1, 32)=4.50, p=.042, η_p ²=.12. The effect was larger in the aided condition, but it was also significant in the unaided condition, F(1, 16)=8.05, MSE=72.93, p=.012, η_p ²=.34. Apart from trends for main effects of method and cue type, all other effects were non-significant, p>.384. Although homographs tended to be less well recalled (and less well learned initially) than non-homographs, SIF did not depend on the nature of the cues.

⁴The same randomised-block order was maintained across phases to reduce variability associated with differential delays.

⁵To reduce error variance a list factor representing the rotation of materials across conditions was also included in the design.

⁶Other, unanalysed responses to the cues for free association included words indicating a third meaning of the homographs (a total of 6 out of 522 responses), words of unknown inspiration (9), and cue repetition (1). On 10 occasions participants failed to respond. These frequencies were distributed approximately evenly across experimental conditions, with the exception of the third-meaning responses; five of the six instances occurred in the unaided suppress condition and are consistent with efforts to reinterpret cues.