Selecting and ignoring the component features of a visual object: A negative priming paradigm

Abstract

Decades of investigation have led to tremendous progress in our understanding of the mechanisms that underlie selective attention to spatial locations and to individual objects. Much less work has been devoted so far to exploring the ability of humans to select the individual features of a multidimensional visual object. Here we report the results of two related experiments in which we used a negative priming procedure to assess whether and under which conditions attention mechanisms can lead to selective processing of the relevant feature of an object (e.g., colour) and/or suppression of the irrelevant features of the same object (e.g., direction of motion or orientation). Results showed that: (1) Individual features of a single object can indeed undergo different processing fates as a result of attention. While one is made available to response selection stages, others are actively blocked. (2) Feature-selective attention most likely operates through a combination of facilitatory and inhibitory mechanisms. (3) In particular, the engagement of inhibitory mechanisms appears to be critically dependent upon the need to resolve response conflict interference between the constituent features of the object. These results are discussed in relation to several ongoing debates concerning the cognitive architecture of attention, including the processing stages at which attentional mechanisms intervene and the types of representation upon which they act.

Acknowledgments

This work was supported by grants to LC from the Italian Government (Ministero dell'Istruzione, Universita’ e Ricerca scientifica, MIUR) and the Human Frontier Science Program (HFSP), and by a joint grant to LC and ACN from the McDonnell-Pew Program in Cognitive Neuroscience. AF was receiving a fellowship covered by the grant from the McDonnell-Pew Program in Cognitive Neuroscience. We wish to thank Giuseppe di Pellegrino and Gianfrancesco Gervasoni, who were involved in the early development of this work. Special thanks to Rebecca Saxe, who gave a critical contribution to performing Experiment 2.

Notes

¹We chose to use one-dimensional, instead of two-dimensional probes, after pilot experiments had shown that reliable negative priming effects could be obtained with probes devoid of any element of response conflict interference.

²We used pairwise t-test comparisons for post hoc analyses throughout, with the Bonferroni correction when required.

³A comparison of the feature and dimension effects for colour probes in terms of error rates revealed a trend for more errors in the former condition, t(13) = 2.10, p=.056.

⁴We are fully confident that this is not due to noise in the present data, since previous and subsequent experiments in our laboratories have revealed only weak (and sometimes negligible) negative priming effects for colour features.

⁵We thank Tram Neill for pointing out to us this alternative interpretation.

⁶There were several methodological differences between Experiments 1 and 2. To a large extent this is due to the fact that Experiment 2 had the additional objective of allowing us the scalp recording of event-related electrical potentials (ERPs) during task performance, and some features of the paradigm were chosen accordingly. Nonetheless, we believe that such methodological differences between the two experiments do not weaken our discussion of the behavioural results of Experiment 2 for the present purposes.

⁷A robust and highly significant negative priming effect for motion has been found in successive replications of this experiment.