Processing of “unattended” threat-related information: Role of emotional content and context

Abstract

Unpleasant, pleasant, or neutral visual scenes served as a context for the presentation of threat-related, positive, and neutral words. On each trial, 2 simultaneous prime words (one foveal, i.e., at fixation, and one parafoveal, i.e., 2.2° apart) appeared for 150 ms, followed by a foveally presented probe word in a lexical decision task. Results showed facilitation in response times for probe threat words when primed by an identical parafoveal word, in comparison with priming by an unrelated parafoveal word, and this effect was enhanced in an emotionally congruent unpleasant context. In contrast, no parafoveal effect appeared for positive words, even in a pleasant context. This reveals parallel processing of threat-related words outside the focus of attention.

Acknowledgments

This research was supported by Grant SEJ2004-420PSIC, from the Spanish Ministry of Education and Science. We are grateful to Jukka Hyona, Juan J. Ortells, and Margaret Dowens for their helpful comments on an earlier version of this article.

Notes

¹'By “unattended” we mean overtly unattended (i.e., that the parafoveal word is not directly fixated), although it is possible that it can be covertly attended to (i.e., seen without being looked at) (see Findlay & Gilchrist, 2003). Typical procedures investigating parafoveal processing involve the presentation of two simultaneous prime words (one foveal, one parafoveal) briefly (150 ms or less), and spatially separated (2° of visual angle or more). Attention is allocated to the foveal, but not to the parafoveal prime: The viewer has to attend to the former (as it is immediately replaced with the foveal probe) and ignore the latter. In such conditions, and taking into account that minimal saccade latency is 150 ms (see Rayner, 1998), it is very unlikely that the eyes can fixate on the parafoveal prime (although the most appropriate research strategy would involve monitoring of eye fixations). This is why it is called the “unattended” prime. Accordingly, parafoveal processing is probably automatic in the sense of being fast and parallel (occurring at the same time as processing of the foveal prime), without overt attention (which is devoted to the foveal prime), unintentional (the goal of the viewer is to process the foveal prime), and unconscious (participants are typically unable to report the parafoveal words).

²We predicted parafoveal positive (rather than negative) priming for our 300 ms SOA presentation on the basis of prior findings regarding the time course of activation of unattended or “ignored” prime words. Prior research has found positive priming between 200 ms and 300 ms, whereas negative priming has emerged later, between 600 ms and 900 ms (e.g., Ortells et al., 2001). This suggests that there is initial automatic activation of the parafoveal information, which is followed by inhibition (as a mechanism to remove distraction caused by the parafoveal stimulus on the attended, foveal stimulus).

³We presented fewer nonword (i.e., negative lexical decision responses) than word trials (i.e., positive responses) to reduce the nonword ratio in order to minimise the involvement of postlexical strategies (see Neely, 1991, for a discussion of this issue).

⁴It may be thought that there is no need to account for these data in terms of the spatial broadening mechanism, and that a temporal speeding mechanism would be sufficient. If this were so, we could expect the same superiority of threat words when presented foveally as when presented parafoveally. Although we have not manipulated the valence and relatedness of the foveal words in this study, Calvo and Castillo (2005) did, using the same 150 ms presentation of foveal primes, followed by an identical or an unrelated probe. These authors found strong foveal priming effects, which were practically identical for the threat and the neutral words (positive priming scores of 99 ms vs. 100 ms, respectively). This is important to rule out the temporal speeding account. It suggests that the parafoveal prime advantage for threat words that we have found in the present study is not simply due to a quicker, but to a broader, perception for these words.

⁵The prime word was the same as the probe word in the related condition, although presented in a different letter type, to keep meaning while reducing perceptual similarity. Our initial intention was to use a semantic priming paradigm, with lexical associates as prime and probe words, but we soon noticed that this was not viable for emotional words. Positive and threat-related word categories are much more restrictive in number of exemplars than the neutral word category. In addition, the emotional categories typically have more specific meanings and nuances, which makes it difficult to find clear semantic associates. This was further complicated by the following constraints: Words should not exceed seven-letter length and should not be infrequent, and length and lexical frequency should be comparable for the three word categories. All these restrictions led us to use an identity-priming paradigm.