Two-object attentional interference depends on attentional set

Transcription

1 International Journal of Psychophysiology 53 (2004) Two-object attentional interference depends on attentional set Maykel López, Valia Rodríguez*, Mitchell Valdés-Sosa Department of Cognitive Neuroscience, Cuban Neuroscience Center, Ave. 25 #15202, esq. 158, Cubanacán, Playa 11600, C. Havana PB 6412, Cuba Received 30 July 2003; received in revised form 10 March 2004; accepted 22 March 2004 Available online 24 May 2004 Abstract In this study, we explored the influence of an irrelevant translational event on the automatic capture of attention to one of two superimposed surfaces defined by transparent motion. The results showed that an irrelevant translation on one surface did not automatically capture the subjects attention if the attentional resources have been endogenously allocated on the other surface. Moreover, the reduction in the motion-onset component of the event-related potential observed in trials where the irrelevant event affected the uncued surface supports the existence of a top-down control of early sensorial processing in this paradigm. This study provides further evidence of the interaction of stimulus-driven and goal-directed mechanisms in the control of visual attention. D 2004 Elsevier B.V. All rights reserved. Keywords: Visual attention; Object; Motion; Attentional set; Stimulus-driven; ERP 1. Introduction In everyday life we use goals and previous knowledge of visual scenes to define a set of information relevant to our current behavior. The perceptual/attentional set biases the subsequent processing of the incoming information in such a way that only events that are behaviorally relevant are easily detected or identified (Bacon and Egeth, 1994; Folk et al., 1992, 1994). This characteristic of attentional control is present in a paradigm developed by Valdes-Sosa et al. (2000) and Pinilla et al. (2001) in which they examined the effect of object-based attention on the early modulation of event-related potentials. * Corresponding author. Tel.: ; fax: address: valia@cneuro.edu.cu (V. Rodríguez). In these studies, subjects were cued by the fixation point (FP) color to selectively attend to one of two imaginary surfaces (one red, one green) induced by transparent motion. The two transparent surfaces can be conceived as two different perceptual groups or objects (Valdes-Sosa et al., 2000) which are segregated on the basis of color and direction of rotation. Brief periods of rectilinear motion (events) interrupted the rotational motion of each surface. The conjunction of distinctive color and motion contributes to preserve the identity of each surface despite changes in its direction of motion. The subjects task consisted of reporting the direction of two such consecutive events that could affect the same (cued) surface on some trials, or different surfaces on other trials. Under these conditions, a behavioral two-object attentional cost was found for the second event in the different surface condition, accompanied by a reliable attentional modulation of /$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi: /j.ijpsycho

2 128 M. López et al. / International Journal of Psychophysiology 53 (2004) the motion-onset event-related potential (ERP). The N200 component related to the second event was relatively suppressed for the uncued but not for the cued surface (Pinilla et al., 2001). This temporal limitation of attention during the discrimination of events occurring close together in time and in different surfaces is known as attentional blink (AB) for motion stimuli (see for a review: Valdes-Sosa et al., 2004). Valdes-Sosa et al. (2000) interpreted the two-object interference as a consequence of a sluggish transition of attention from one surface to the other. However, the selection of the initially attended object could be accomplished by either the use of the information contained in the cue (dubbed as endogenous cue for instructing subjects to attend a specific set of dots, instruction that gives rise to an endogenous allocation of attention) or by exogenous attentional capture (i.e. the first event would grab attention). The relative contribution of each type of mechanism was not explored in these initial reports. In a recent study, Reynolds et al. (2003) replicated the findings described above, but questioned the role of the cue in the initial endogenous allocation of attention in this paradigm. These authors obtained a two-surface cost even when their FP was uninformative (gray on all trials). They concluded that the first event exogenously captured attention, and that endogenous cueing played little role in Valdes-Sosa et al. s experiments. Although the findings of Reynolds et al. convincingly demonstrate that purely exogenous capture of attention can lead to a two-object cost, they do not rule out the participation of endogenous cueing in the previous experiments with transparent surfaces. In fact, there is an important difference in procedure between these studies. In the original paradigm, subjects were asked to focus attention on one surface at the beginning of each trial, whereas in Reynolds study, subjects had to divide attention between the two surfaces. Previous work has shown that exogenous attentional capture is enabled by some, and precluded by other, attentional settings. For example, in visual search tasks where singletons (one item of the search display with a unique, distinctive feature) are used, it has been observed that a color singleton will capture attention automatically only if the subject is searching for singletons defined by another feature dimension like shape (singleton search mode) (Bacon and Egeth, 1994; Folk et al., 1992, 1994). Motion will capture attention automatically only when it is relevant to the subjects task or when it defines a new object (Hillstrom and Yantis, 1994; Abrams and Christ, 2003). Therefore, it is possible that in the experiments carried out by Valdes-Sosa s group, the cue helped to endogenously allocate attention to one surface, but that there is exclusively exogenous capture when attention must be divided between the surfaces (as in Reynolds study). This idea can be tested by symbolically cueing the subjects to attend to one surface and measuring the degree of attentional capture produced by an event on the other surface. If endogenous cueing plays no role, then automatic attentional capture would occur independently of where the subjects attention was endogenously directed to. In the present study, we show that the occurrence of an irrelevant translational event does not automatically capture the subjects attention if the attentional resources have been endogenously placed elsewhere. 2. Materials and methods The experimental methods are only outlined here, for more details, readers are referred to previous articles (Pinilla et al., 2001; Valdes-Sosa et al., 2000). Ten subjects (five males) participated as volunteers in the experiments. Ages ranged between 22 and 28 years. All subjects had normal or corrected-tonormal vision and no history of neurological or psychiatric disorders. Due to the difficulty of the task, potential subjects first practiced and were replaced if accurate performance (above 75% correct responses) was not obtained. The stimuli consisted of two interspersed sets of moving dots, one colored red and the other colored in an isoluminant green (Fig. 1). Two experiments were carried out by the subjects. In both, after a baseline period in which the two sets of dots rotated rigidly and in opposite sense around a fixation point (FP) during 750 ms, a first translational event (T1) lasting 150 ms affected one of the surfaces while the other continued to rotate. After this, both surfaces rotated in the baseline pattern for 350 ms until a second 150 ms translational event (T2) was presented. The trial ended with an additional 700 ms rotation period. In order to

3 M. López et al. / International Journal of Psychophysiology 53 (2004) Fig. 1. Sequence of stimulus events in a trial. The background was actually black, and dots were either red or green isoluminant with the red (represented as black and gray). (A) Events in same-surface trials: both events, T1 and T2 affected the same surface. (B) Events in differentsurface trials: T2 affected a different surface than T1. In Experiment 2, the FP was colored red or green warning the subject where T2 was going to occur. However, in Experiment 1, the FP (gray) was uninformative. make T1 irrelevant, it always moved to the upward direction and no response to it was required. However, the direction of T2 was always different and randomly selected from eight possible alternatives. The experiments took place in a room with dim illumination. Subjects were instructed to maintain fixation and minimize body movement and eyeblinks during recording blocks. The task of the subjects was to discriminate the direction of T2 ignoring the previous T1. A gray or a colored (red or green) FP was used depending on the experiment. In Experiment 1, a gray FP was used, providing the subjects no information about which surface was going to be affected by T2. Due to the lack of cueing in this experiment, attention had to be divided between both surfaces in order to succeed in T2 discrimination. However in Experiment 2, the FP color warned the subjects where T2 was going to occur, on the red or on the green surface. Here, the color of the FP worked as an endogenous cue that enabled selective attention to the surface with the same color as the FP. Each experiment was performed in separate sessions carried out on different days. Sessions were comprised by two blocks of 335 trials each. With the aim to make ignoring T1 easier, its occurrence on the red or on the green surface was fixed within a block: in one block, it occurred on the red surface and the other on the green one. The presentation order of blocks and experiments was counterbalanced across the subjects. Two types of trials were used and randomly presented: same-surface (where T1 and T2 occurred in the same surface) and differentsurface (where T2 occurred in a different surface than T1). The surface to be attended and the type of trial were always presented in a random order. Subjects reported the direction of T2 on the numericalpad of the computer keyboard. Incorrect responses were signaled by a 500 ms beep on the computer loudspeaker. Electrophysiological data acquisition was carried out on a MEDICID 3E system (Neuronic SA). Disk electrodes (Ag/AgCl) were placed in five active derivations (O1, O2, T5, T6, Oz) of the international system 10/20. All active electrodes were referred to linked earlobes. Inter-electrode impedance was always kept below 5 kv. Two bipolar derivations were used

4 130 M. López et al. / International Journal of Psychophysiology 53 (2004) to record the electro-oculogram (EOG), with one electrode placed just lateral to the external canthi and another 1 cm above the right eye for horizontal and vertical movements, respectively. The signals were filtered between 0.05 and 30 Hz (3 db down). Additionally, a notch filter with a peak at the power line (60 Hz) frequency was used. In each trial, marks corresponding to stimulus events (onset of T1) were co-registered with the amplified and digitized electroencephalogram (EEG) (12-bit resolution), which was sampled at a rate of 250 Hz and stored on magnetic disk for off-line analysis. The recorded EEG was windowed with a prestimulus baseline of 100 ms before T1 onset, and a 900 ms post-stimulus epoch. EEG segments with artifacts or excessive activity in the EOG were rejected. This eliminated about 5 25% of all stimulus events, which resulted in individual ERPs based on the average of around 180 events. Grand average ERPs were calculated for all subjects for each site and condition. All data points were corrected (prior to plotting or measurement) by subtracting the average pre-stimulus amplitude value. The percentage of correct responses were adjusted to compensate for guessing (guessing-level of 0.125). A two-way repeated measures ANOVA (rm-anova) was carried out with the psychophysical data using CUE (colored versus gray FP) and TRIAL-TYPE (same- versus different-surface) as main factors. Since the color of the dots did not produce any significant effect in subsequent analysis, data were collapsed over this factor in the ERP and behavioral data. The effect of endogenous cueing on the ERP components were tested with rm-anova and a t- test corrected with non-parametric permutation techniques (Blair and Karniski, 1993; Galan et al., 1997). For the rm-anova time windows centered on the peak of the N200 elicited by T1 and T2 were measured. We named these two N200 as N200-T1 and N200-T2, respectively. In the case of N200-T1 the time window ranged from 170 to 330 ms (with respect to T1 presentation), in the case of N200-T2 the window ranged from 631 to 791 ms (with respect to T1 presentation too). A one-way rm-anova was performed with TRIAL-TYPE (same- versus different-surface) as main factor in each experiment. Besides, differences between same- and differentsurface trials of both experiments were also explored. A one-way rm-anova with CUE (colored versus gray FP) as main factor was performed for each type of trial. 3. Results Fig. 2 shows the proportion of correct responses for T2 in colored and gray FP experiments, as a function of whether the two events affected the same or different surfaces. As can be observed in the figure, the subjects performance for T2 was better when they Fig. 2. Means and standard errors of correct responses for T2 as a function of whether the two events affected the same or different surfaces. Each line represents a different experiment. Responses were more accurate in Experiment 2 (colored FP) for both type of trial. However, subjects performance dropped in Experiment 1 (gray FP) in different-surface trials.

5 M. López et al. / International Journal of Psychophysiology 53 (2004) knew in advance which surface was going to be affected. In the colored FP experiment (solid line in Fig. 2), T2 was discriminated accurately in both sameand different-surface trials. However, when the FP was gray (dashed line in Fig. 2), there was a deterioration in performance if T2 occurred on a different surface than the one affected by T1. These effects were test in rm-anova that showed highly significant effects of CUE ( F(1,8) = 39.9, p < ) and TRIAL-TYPE ( F(1,8) = 17.1, p < ). The interaction between these factors was also significant ( F(1,8) = 5.1, p < 0.05). In planned comparison no trial-type effect was found in Experiment 2 (colored FP). The presence of the endogenous cue made it possible to discriminate T2 regardless of whether it affected the same or a different surface than T1. However, when subjects did not have advanced information on the surface which would be affected by T2 (Experiment 1), performance for this event was significantly different depending on trial type, with same- more accurate than different-surface trial ( F(1,8) = 11.6, p < 0.01). The generally better scores obtained in Experiment 2 compared to Experiment 1 were also tested using planned comparisons. This analysis showed significant differences between same- ( F(1,8) = 11.7, p < 0.01) and different-surface trials ( F(1,8) = 20.1, p < 0.002) of both experiments. The grand average ERPs obtained from T5 and T6 derivations are displayed in Fig. 3. The ERPs for the different conditions same- versus different-surface trials in Experiments 1 and 2 are shown superimposed. Two negative peaks corresponding to the occurrence of T1 and T2 were present in the individual ERPs in all participants. A modulation in the amplitude of N200-T1 depending on FP color was found. As expected the amplitude of N200-T1 was smaller in different- than in same-surface trials when an endogenous cue informed in advance the subjects on the position of T2. The rm-anova showed that the presence of a cue (Experiment 2) significantly influenced ( F(1,9) = 6.1, p < 0.03) the amplitude of N200-T1. The component was smaller in different- than in same-surface trials Fig. 3. Grand average ERP from Experiment 1 (gray FP) and Experiment 2 (colored FP). Responses to same- and different-surface trials are overlaid. The time axis is referred to the onset of the first event. The two arrows indicate times when events (T1 and T2) were presented. Observed significance level obtained with permutation test for the difference between same- and different-surface trials are also shown for each experiment.

6 132 M. López et al. / International Journal of Psychophysiology 53 (2004) (Fig. 4A). However, when no cue was available on the T2 surface (Experiment 1), no differences were found in N200-T1 amplitude between same- and differentsurface trials. As can be observed in Fig. 3, the negativity provoked by the second translation, the component N200-T2, was larger in trials were T2 occurred on the same surface as T1 in both experiments. However, note that this component seems to be larger in Experiment 2 (colored FP) than in Experiment 1 (gray FP). The rm-anova performed on the amplitude of N200-T2 showed significant differences in the amplitude of N200-T2 between same- and different-surface trials in colored FP ( F(1,9) = 16.8, p < 0.002) and gray FP ( F(1,9) = 17.2, p < 0.002) experiments. That is, in both experiments N200-T2 was more negative in same- than in different-surface trials (Fig. 4B). This analysis also showed that ERPs obtained in Experiment 2 are indeed significantly more negative than in Experiment 1 for both types of trial (same-surface ( F(1,9) = 7.6, p < 0.02) and different-surface ( F(1,9) = 13.4, p < 0.005)). The difference in the N200-T1 region (from 220 to 275 ms) between neural responses related to the two types of trials was found significant ( p < 0.03) in the t- test corrected with permutation techniques (Fig. 3) for Experiment 2 (colored FP) but not for Experiment 1 (gray FP). Significant differences ( p < 0.05) were also found with the same technique in the N200-T2 region ranging from 655 to 715 ms of both experiments. 4. Discussion The results obtained in the present study indicate that the prior allocation of attentional resources to a given object by a cue attenuates the response to uninformative and irrelevant events taking place in the object that has not been cued. However, even when this suppression is not absolute, it does not interfered with the discrimination of a subsequent T2 affecting another surface. This result supports and extends Hillstrom and Yantis (1994) finding that a motion event per se does not capture attention if it is irrelevant for the task. However, when no cue was given on T2 surface, subjects had to divide attention between surfaces to optimize its discrimination. Sharing attention between the two surfaces made it impossible to filter out the first event causing a drop in T2 performance. Several authors have argued that stimulus-driven capture depends on attentional settings (Bacon and Egeth, 1994; Ludwig and Gilchrist, 2002; Yantis and Jonides, 1990). This implies that, depending on the experimental conditions, exogenous capture of attention is subject to top-down control. Moreover, when attention was divided between surfaces due to the lack of cue there was no difference Fig. 4. (A) Means and standard errors of N200 amplitude for T1 presentation. Note that the N200 was smaller in different-surface trials when a cue was provided. However, the amplitude of this component is similar for same- and different-surface trials in the absent of the cue. (B) Means and standard errors of N200 amplitude for T2 presentation. The component was more negative in same- than in different-surface trials irrespective of the cue presence.

7 M. López et al. / International Journal of Psychophysiology 53 (2004) in the N200-T1 amplitude between same- and different-surface trials. The difference between these conditions was, however, big when it was possible to focus attention in one surface. The reduced N200-T1 (observed in two-surface trials when an endogenous cue was provided) supports the existence of a filtering process for irrelevant events in early visual areas when attention is previously allocated elsewhere. This filtering process could have been favored by the occurrence of T1 in a fixed surface during a block. However, the processing of the irrelevant T1 that affects the cued surface cannot be prevented and a well-defined N200- T1 component is observed in this case, confirming the conclusion stated in the paragraph above. On the other hand, the fact that the negativity produced by the second event (the N200-T2 component) was larger in same- than in different-surface trials of both experiments could be an evidence of summation of the neural activation produced by the first and the second event in these trials. Moreover, the presence of a larger N200-T2 in same-surface trials after endogenous cueing compared to when this cue was absent, suggest an interaction between stimulus-driven (the occurrence of a previous translation) and goal-directed (the endogenous cue) mechanisms (Corbetta and Shulman, 2002; Pashler et al., 2001). Actually, some studies have reported increases in the baseline activity of neurons (Kastner et al., 1999; Luck et al., 1997; Ress et al., 2000; Shulman et al., 1999) during attention tasks and have proposed that it could potentiate an enhancement in stimulus-evoked responses (Ress et al., 2000). It is possible that the activation caused by the cue in the Experiment 2 plus those produced by the occurrence of the first translational event facilitated the response of the neural population to the second event causing a larger ERP. However, it cannot be ruled out that the small reduction observed in the amplitude of N200-T2 in different-surface trials compared to same-surface trials when the cue is available could also be influenced by an exogenous modulation produced by T1 that attenuates the endogenous effect on the component. But if this exogenous attenuation does exist, it is not enough to behaviorally produce a two-object interference. Nevertheless, an experimental design that controls the possible exogenous driving of T1 from of the endogenous effects produced by the cue should be carried out. Taken together, the results obtained in this study suggest that the occurrence of a translational event in one surface is not sufficient to automatically capture attention when it has been endogenously allocated in another surface. The capacity of a motion event for biasing the sensorial processing in favor of one stimulus depends on the pre-established attentional set. It is not the translation per se but its occurrence in the selected surface which influenced the amplitude of the N200 component in this study. Visual selection is a process that, depending on the context, differentially weighs stimulus-driven and goal-directed influences. References Abrams, R.A., Christ, S.E., Motion onset captures attention. Psychol. Sci. 14 (5), Bacon, W.F., Egeth, H.E., Overriding stimulus-driven attentional capture. Percept. Psychophys. 55, Blair, R.C., Karniski, W., An alternative method for significance testing of waveform difference potentials. Psychophysiology 30, Corbetta, M., Shulman, G.L., Control of goal-directed and stimulus-driven attention in the brain. Nat. Rev. Neurosci. 3, Folk, C.L., Remington, R.W., Johnston, J.C., Involuntary covert orienting is contingent on attentional control settings. J. Exp. Psychol. Hum. Percept. Perform. 18, Folk, C.L., Remington, R.W., Wright, J.H., The structure of attentional control: contingent attentional capture by apparent motion, abrupt onset, and color. J. Exp. Psychol. Hum. Percept. Perform. 20, Galan, L., Biscay, R., Rodriguez, J.L., Testing topographic differences between event related brain potentials by using nonparametric of permutation tests. Electroencephalogr. Clin. Neurophysiol. 102, Hillstrom, A.P., Yantis, S., Visual motion and attentional capture. Percept. Psychophys. 55, Kastner, S., Pinsk, M., Desimone, R., Ungerleider, L., Increased activity in human visual cortex during directed attention in the absence of visual stimulation. Neuron 22, Luck, S.J., Chelazzi, L., Hillyard, S.A., Desimone, R., Neural mechanisms of spatial selective attention in areas V1, V2 and V4 of macaque visual cortex. J. Neurophysiol. 77, Ludwig, C.J., Gilchrist, I.D., Stimulus-driven and goal-driven control over visual selection. J. Exp. Psychol. Hum. Percept. Perform. 28, Pashler, H., Johnston, J.C., Ruthruff, E., Attention and performance. Annu. Rev. Psychol. 52, Pinilla, T., Cobo, A., Torres, K., Valdes-Sosa, M., Attentional shifts between surfaces: effects on detection and early brain potentials. Vis. Res. 41,

8 134 M. López et al. / International Journal of Psychophysiology 53 (2004) Ress, D., Backus, B.T., Heeger, D.J., Activity in primary visual cortex predicts performance in a visual detection task. Nat. Neurosci. 3, Reynolds, J.H., Alborzian, S., Stoner, G.R., Exogenously cued attention triggers competitive selection of surfaces. Vis. Res. 43, Shulman, G.L., Ollinger, J.M., Akbudak, E., Conturo, T.E., Snyder, A.Z., Petersen, S.E., Corbetta, M., Areas involved in encoding and applying directional expectations to moving objects. J. Neurosci. 19, Valdes-Sosa, M., Cobo, A., Pinilla, T., Attention to object files defined by transparent motion. J. Exp. Psychol. Hum. Percept. Perform. 26, Valdes-Sosa, M., Bobes, M.A., Rodríguez, V., Acosta, Y., Pérez, A., Iglesia, J., Borrego, M., The influence of scene organization on attention: psychophysics and electrophysiology. In: Kanwisher, N., Duncan, J. (Eds.), Attention and Performance XX: Functional Newimaging of Visual Cognition. Part 3, Chapter 16 Oxford University Press, Oxford. Yantis, S., Jonides, J., Abrupt visual onsets and selective attention: voluntary versus automatic allocation. J. Exp. Psychol. Hum. Percept. Perform. 16,