Affective significance enhances covert attention: Roles of anxiety and word familiarity

Abstract

To investigate the processing of emotional words by covert attention, threat-related, positive, and neutral word primes were presented parafoveally (2.2° away from fixation) for 150 ms, under gaze-contingent foveal masking, to prevent eye fixations. The primes were followed by a probe word in a lexical-decision task. In Experiment 1, results showed a parafoveal threat–anxiety superiority: Parafoveal prime threat words facilitated responses to probe threat words for high-anxiety individuals, in comparison with neutral and positive words, and relative to low-anxiety individuals. This reveals an advantage in threat processing by covert attention, without differences in overt attention. However, anxiety was also associated with greater familiarity with threat words, and the parafoveal priming effects were significantly reduced when familiarity was covaried out. To further examine the role of word knowledge, in Experiment 2, vocabulary and word familiarity were equated for low- and high-anxiety groups. In these conditions, the parafoveal threat–anxiety advantage disappeared. This suggests that the enhanced covert-attention effect depends on familiarity with words.

Keywords:

• Anxiety
• Attention
• Eye movements
• Parafoveal
• Priming
• Threat
• Word familiarity

Acknowledgments

This research was supported by Grant SEJ2004–00420PSIC, from the Spanish Ministry of Education and Science. We are grateful to Lauri Nummenmaa for his assistance in implementing the gaze-contingent masking technique.

Notes

¹ In practical terms, we found difficulty in implementing a semantic paradigm for emotional words. First, positive and threat-related word categories are much more restrictive in number of exemplars than the neutral category. Second, many emotional words involve specific meanings, which prevents them from being interchanged with other words, as semantic associates. Third, there were some methodological constraints: Words should not exceed seven-letter length and should not be infrequent, and length and frequency should be comparable for the three word categories, which considerably reduces the number of stimuli.

² As noted by a reviewer, the means in Table 2 and Figure 3 go in a direction consistent with a residual effect of anxiety on threat priming (even though the critical interaction was not significant) in Experiment 2. Furthermore, a tendency remained for the high-anxiety group to exhibit threat priming after word familiarity was factored out in Experiment 1 (ps = .085 and .11). Accordingly, given the importance of determining the extent to which word familiarity is responsible for the threat–anxiety priming effects, a more stringent, hypothesis-driven analysis was conducted on the data from Experiment 2. When the threat and the neutral trials were considered alone, both the anxiety by valence by relatedness interaction and the four-way interaction remained nonsignificant (ps = .51 and .33, respectively). Also, comparisons between the anxiety groups on the priming scores for threat and neutral trials separately yielded nonsignificant differences (ps = .54 and .75, respectively). This confirms that the threat–anxiety priming effect is greatly (though maybe not completely) dependent on word familiarity.