Irrelevant digits affect feature-based attention depending on the overlap of neural circuits

Transcription

1 Cognitive Brain Research 12 (2001) locate/ bres Research report Irrelevant digits affect feature-based attention depending on the overlap of neural circuits Wim Fias *, Johan Lauwereyns, Jan Lammertyn a, b a a Department of Experimental Psychology, Ghent University, Henri Dunantlaan 2, B-9000 Ghent, Belgium b Department of Physiology, Juntendo University, Japan Accepted 13 July 2001 Abstract Feature-based attention was investigated by examining the effect of irrelevant information on the processing of relevant information. In all experiments, irrelevant information consisted of digits whose semantic information is known to be processed in parietal areas. Between experiments we varied the degree of parietal involvement in the processing of the relevant feature. The influence of the irrelevant digit on the binary manual response task on the relevant feature was measured by the SNARC effect, a spatial numerical association of response codes demonstrating faster left than right hand responses for small numbers and faster right than left hand responses for large numbers. When processing of the relevant feature depended on parietal cortex, as is the case for orientation processing (exps. 1 and 4), there was an effect of the digit s semantic value on response times. Conversely, there was no effect of the irrelevant digit on the processing of color (exps. 2 and 3) or shape (exp. 5), which rely only minimally on parietal resources. After ruling out alternative explanations we conclude that the efficiency of feature-based attention is determined by the degree of neural overlap of structures dedicated to process relevant and irrelevant information Elsevier Science B.V. All rights reserved. Theme: Neural basis of behavior Topic: Cognition Keywords: Feature-based attention; Shape; Color; Orientation; Number processing; Ventral/ dorsal 1. Introduction feature-based attention operates as a localized neural adaptation to the task at hand. In line with this view, it is Efficient behavior requires selection of information. In possible that the quality of feature-based selection depends the domain of vision, the information acting upon our on the extent to which relevant and irrelevant information retinas is too abundant to be processed all to the same are active in the same neural structures. That is to say, degree during the conversion of sensory input into goal- interference from irrelevant information on feature-based oriented behavioral output. Attentional mechanisms pro- attention would be stronger if the relevant and the irrelevide a means to give privilege to a subset of the available vant feature are processed by the same neural structures information [25,43]. Selection can be accomplished on the than if both features pass through different neural strucbasis of a particular area in space [23,35], a certain object tures. [14,44,46] or a particular visual feature [8,9,26]. In this To test this hypothesis, we devised a behavioral paper we focus on feature-based attention. paradigm in which we varied the relevant information Single-unit studies [29,41] and functional imaging while keeping irrelevant information constant. As irrelestudies [6,8,9,34] indicate that feature-based attention vant information we used digits. Functional imaging (e.g. modulates visual processing in specific areas of neocortex Refs. [7,33]), electrode recordings [1] and patient studies depending on which feature is being attended. Thus (e.g. Ref. [12]) convergently demonstrate the involvement of superior and inferior parietal areas in the semantic *Corresponding author. Tel.: ; fax: processing of the numerical magnitude of digits. The address: wim.fias@rug.ac.be (W. Fias). pre-semantic physical identity of digits is processed in / 01/ $ see front matter 2001 Elsevier Science B.V. All rights reserved. PII: S (01)

2 416 W. Fias et al. / Cognitive Brain Research 12 (2001) inferior extrastriate cortical areas [2,33]. Behavioral studies 2.2. Stimuli suggest that magnitude-related information is autonomously activated, as witnessed by effects of number magnitude The numbers used ranged from 0 to 9 (except exp. 5, 1 on processing times in tasks for which a mere visual to 9) and were presented centrally on a black background analysis of the digit suffices [10]. As relevant information as Arabic digits in Borland C s simplex font (VGA card in we used either contour-based 2D shape, color, or orienta- graphics mode). The digit subtended an area of tion. These features are processed to a different degree by mm. In the color the digit was colored in standard red or parietal areas. Parietal involvement in the processing of green (exp. 2) and lightcyan and cyan (exp. 3), as shape and color is very restricted, whereas it is more standardly defined in Borland C s library. In all other extensive in the case of orientation processing [15 experiments the digits were presented in white with a 17,19,32,37,38,45], not only in visually-guided motor tasks visual stimulus superimposed in the centre. In exp. 1 this [30,32] but also in associative visuo-motor transformation stimulus was a triangle pointing upward or downward tasks like the two-choice manual response task adopted (subtending mm). In exp. 4 an oriented line here [17,19,37,38]. Accordingly, we expected that seman- segment (1831 mm) was superimposed on the digit and in tic interference from number would be observed with exp. 5 a circle or a square (18318 mm). attention to orientation, but not or to a lesser degree with attention to color or shape Procedure Semantic processing of digits can be evaluated in a binary manual response task. Dehaene et al. [11] first Participants had to press one of two response keys, showed an association between numerical value and the depending on the relevant feature (exp. 1: triangle pointing side of the response and labeled this effect the SNARC upward or downward; exp. 2: red or green; exp. 3: light or effect (spatial numerical association of response codes). dark cyan; exp. 4: horizontal or vertical line; exp. 5: circle Participants were asked to press one of two keys in or square). Key assignment was counterbalanced across response to an even number and the other key to an odd participants. Before the actual experiment the participants number. Smaller numbers were responded to faster with took part in a training session, consisting of 20 trials with the left hand than with the right hand. The reverse letters instead of numbers. In the training session an happened for larger numbers. Additional control experi- auditory feedback buzz was given when the participant ments confirmed that the SNARC effect originates from pressed the wrong key. In the actual test session, each accessing a number s semantic magnitude representation, number was presented a fixed number of times (32 in exps. which is conceived of as a position on a left-to-right 1, 2 and 5; 24 in exps. 3 and 4): half of the times with the oriented mental number line [4,11] such that there is one relevant feature, the other half of the times with the congruity between small numbers and left-hand responses other relevant feature. The whole set of trials was preand between large numbers and right-hand responses. sented in randomized order with a different randomization Therefore, we used the SNARC effect as a marker for for each subject. semantic access. By having subjects direct attention away Each trial started with the symbol [ as fixation point from the digit and asking them to perform a two-choice (18 mm350 mm) presented at the center of the screen for manual response task, we can evaluate the extent to which 1000 ms. The subjects were asked to fixate this point, but the unattended digit is processed as a function of the kind eye position was not monitored. Thereafter, the screen was of processing being performed attentively. erased and immediately followed by the stimulus. The stimulus remained on until a response was made, which was registered to the nearest ms from stimulus onset. A 2. Materials and methods blank screen followed for 500 ms, after which the next trial started. The two response buttons were connected to a 2.1. Participants PC-compatible Pentium computer, were placed at a comfortable position in front of the subject and were separated Sixty-eight Dutch-speaking subjects participated in the by 30 cm. The eye screen distance was plus or minus 70 study. All participants had normal or corrected vision, and cm. Both speed and accuracy were stressed in the instrucreported to be neuropsychologically healthy. Experiments tions, and the interval of numbers used was explicitly 1 2 and 3 4 were each tested on the same sample (each mentioned. 24 participants), in which case the order of experiments was counterbalanced. Subsamples were composed com Data analysis parably in terms of age (exps. 1 2: 19.3; exps. 3 4: 18.3; exp. 5: 24.2), sex (number of male participants: exps. 1 2: Trials with an RT shorter than 150 ms or longer than 7; exps. 3 4: 4; exp 5: 11), handedness (number of 1000 ms were discarded from all analyses. The cut-off right-handed participants: exps. 1 2: 20; exps. 3 4: 20; value of 1000 ms fell well outside the grand average plus exp. 5: 16) and education (all participants were students or three standard deviations. This way, less than 1.5% of the recent graduates in psychology). data are excluded.

3 W. Fias et al. / Cognitive Brain Research 12 (2001) The presence of a SNARC effect was evaluated by can be easily segregated from the digit background, in means of a regression analysis of repeated measures data terms of Gestalt laws on the basis of good continuation and as described by Lorch and Myers [28]. We reported the closure. Lesions of inferior parietal cortex have been advantages of this analysis elsewhere [18]. In a first step, shown to impair the discrimination of rotated shapes in the for each subject the median RT of the correct responses monkey [15,45]. Single-cell recordings reported orientation was computed for each number, separately for left and selective cells in the anterior intraparietal area [32]. right responses. On the basis of these medians, differences Functional brain imaging demonstrated the involvement of in RT (drts) were computed by subtracting the median parietal areas in the processing of spatial features like RT for the left hand from the median RT for the right orientation [16,17,19,37,38]. hand. If there is an association between response side and Error rate averaged over subjects was 3.0% (with a number magnitude, we expect a negative correlation maximum of 6.1%). There was no speed accuracy tradebetween number magnitude and drt: relatively small off, as indicated by the absence of a negative correlation numbers should elicit faster left responses, resulting in between RT and number of errors, computed over 20 data positive drts, whereas relatively large numbers should couples (10 numbers, separated for left and right reelicit faster right responses, and thus negative drts. sponses), r510.30; n520; P,0.30. Mean RTs of correct Therefore, in a second step, a regression equation was responses with the digits 0 to 9 as irrelevant information computed per subject with number magnitude as predictor were, respectively, 453, 452, 457, 467, 468, 465, 454, 455, variable. In a third step t-tests were performed to test 445 and 460 ms. whether the regression weights of the group deviated The regression analysis of repeated measures data significantly from zero. Since only a negative slope is revealed the following equation (presented in Fig. 1): expected theoretically, all reported P-values are one-tailed. drt magnitude. 3. Results The significant contribution of number magnitude to the pattern of left hand right hand differences (t(23)522.91, 3.1. Experiment 1: orientation, triangle pointing S.D.53.4, P,0.005) reflects a reliable SNARC effect upwards or downwards (with a negative slope observed in 17 participants) and shows that the unattended digits were processed semanti- Participants had to determine the orientation of a cally. As Brysbaert [4] has argued, the semantic coding of triangle superimposed on a digit (ranging from 0 to 9). the number zero might be different than the semantic Participants had to indicate whether the triangle was coding of other numbers. To make sure that this cannot pointing upwards or downwards. A triangle was chosen as bias our results, we also provide regression equations task relevant feature because it is perceptually salient and computed on the numbers 1 to 9. This reveals a nearly Fig. 1. Differences in RT (drt) between right and left hand responses (right left) as a function of the irrelevant digit. Squares indicate the observed drts. The continuous line depicts the predicted drts on the basis of the regression analysis.

4 418 W. Fias et al. / Cognitive Brain Research 12 (2001) identical result: drt magnitude (t(23)5 drt magnitude , P,0.05). The fact that there is a majority of negative drts does Magnitude was not reliably different from zero (t(23)5 not obstruct the acceptance of a SNARC effect as it is 0.44; S.D.56, P,0.3, with a negative slope in 10 participhighly likely that this is a result of the fact that a large ants). Omission of zero from the regression analyses majority of the participants was right-handed, resulting in reveals a similar equation: drt magnitude overall faster right-hand responses. Then, the SNARC [t(23)50.6; P,0.25). The direct comparison of the mageffect shows how this right-hand advantage is modulated nitude slopes in experiments 1 and 2 revealed a reliably by the magnitude of the presented number. more negative slope in the orientation task than in the color task (t(23)521.86; P,0.05). Thus, whereas a 3.2. Experiment 2: color, red or green reliable SNARC effect was obtained when orientation was the relevant feature, the SNARC effect completely dis- In experiment 2 the relevant feature was changed to appeared when subjects attended to color. color. As parietal cortex is not involved in the processing of color [5,47] but relies primarily on structures in the 3.3. Experiment 3: color, light cyan and dark cyan ventral stream, less interference from the irrelevant digits is expected if participants respond to color as the relevant Average RTs in experiment 2 were considerably shorter feature. In fact, luminance was not controlled. For the than in the orientation task. Possibly, latencies were simply present purposes, however, the difference between color too short for the number semantic system to become and luminance is unimportant as the neural correlates of sufficiently activated to affect performance in the color attentive selection for color and luminance have been task. In order to evaluate this alternative interpretation, a shown to be essentially the same [31]. new color/ luminance discrimination experiment was car- Error rate averaged over subjects was 2.5% (with a ried out, using less discriminable colors/ luminances (light maximum of 7.7%). There was no speed accuracy trade- and dark cyan). off, as indicated by the absence of a negative correlation Error rate averaged over subjects was 4.5% (with a between RT and number of errors, computed over 20 data maximum of 12.9%). There was no speed accuracy tradecouples (10 numbers, separated for left and right re- off, as indicated by the absence of a negative correlation sponses), r520.02; n520; P,1. Mean RTs of correct between RT and number of errors, computed over 20 data responses with the digits 0 to 9 were, respectively, 385, couples (10 numbers, separated for left and right re- 386, 385, 388, 387, 376, 375, 386, 378 and 380 ms. sponses), r50.47, n520; P,0.05. Mean RTs of correct The following equation was obtained and is presented in responses with the digits 0 to 9 were, respectively, 499, Fig. 2: 513, 505, 492, 494, 492, 481, 500, 489 and 489 ms. Fig. 2. Differences in RT (drt) between right and left hand responses (right left) as a function of the irrelevant digit in colour discrimination (red or green). Squares indicate the observed drts. The continuous line depicts the predicted drts on the basis of the regression analysis.

5 W. Fias et al. / Cognitive Brain Research 12 (2001) Even when the colors were chosen such that the average ment 3 with the same set of participants, the participants of latency was raised considerably (from 382 to 492 ms) up experiment 3 were subjected to an orientation task with to a level at which a SNARC effect was obtained in lines instead of triangles. The line could be either vertical experiment 1 (457 ms), there was no indication of an effect or horizontal. An oriented line rather than a triangle was of semantic processing of the digit. The following equation used as a target to evaluate generality. was obtained and is presented in Fig. 3: Technical failure caused data loss for one participant. drt magnitude. Error rate averaged over subjects was 3.5% (with a maximum of 13.3%). There was no speed accuracy tradeoff, as indicated by the absence of a negative correlation Magnitude tended to be positive, the reverse of the between RT and number of errors, computed over 20 data SNARC effect, but was not reliably different from zero couples (10 numbers, separated for left and right re- (t(23)51.5; S.D.56.8; P,0.1, with eight participants sponses), r50.16; n520; P,1. Mean RTs of correct exhibiting a negative slope). The same was true for the responses with the digits 0 to 9 were, respectively, 468, regressions computed with zero excluded: drt5 491, 479, 475, 509, 471, 468, 485, 466 and 476 ms magnitude (t(23)51.58; P,0.1). Inspection Regression equations were computed in order to test for of the drts for the individual numbers suggests a categori- the presence of a SNARC effect. The following equation cal effect of number magnitude. It has been shown that was obtained and is presented in Fig. 4: when precise numerical magnitude is irrelevant to the task, a stimulus can automatically be classified as small or large drt magnitude. without coding the precise numerical value [42]. This Number magnitude was reliably activated, as witnessed crude small large classification has been argued to be by a significant SNARC effect (t(22)523.59; S.D.55.0; essentially different from refined numerical coding [20,42]. P,0.001 with a negative slope in 18 participants). Without Further research is needed to test whether the tendency zero, the equation was drt magnitude observed here is a systematic effect and, if so, to explain (t(22)523.67; P,0.001). The direct comparison of the why crude magnitude information is associated with spatial magnitude slopes from experiments 3 and 4 reveals a coordinates of response codes, opposite to refined numeri- reliably more negative slope in the line orientation experi- ment (t(22)523.27; P,0.01), whereas RTs between the two tasks did not differ (t(22)50.76; P,1). cal magnitude Experiment 4: orientation, horizontal or vertical line segment 3.5. Experiment 5: shape, circle or square To enable a direct comparison of the results of experi- So far, the SNARC effect was present in those situations Fig. 3. Differences in RT (drt) of right and left hand responses (right left) as a function of the irrelevant digit in colour discrimination (light cyan or dark cyan). Squares indicate the observed drts. The continuous line depicts the predicted drts on the basis of the regression analysis.

6 420 W. Fias et al. / Cognitive Brain Research 12 (2001) Fig. 4. Differences in RT (drt) beetween right and left hand responses (right left) as a function of the irrelevant digit in orientation discrimination (horizonal and vertical line). Squares indicate the observed drts. The continuous line depicts the predicted drts on the basis of the regression analysis. where a figure (triangle or line) had to be separated from 4. Discussion the background. No figure ground segregation was required in the color experiments. To judge the possibility In a series of experiments we showed that interference that the SNARC effect results from this figure ground from irrelevant information on feature-based attention is segregation process rather than from the overlapping stronger if the relevant feature is processed by the same parietal neural structures between orientation and digit neural structures as the irrelevant feature. Irrelevant inprocessing, the following experiment was set up. A figure formation was the same in all experiments and consisted of was superimposed on the digit background, but in a digits. Digits are processed semantically in the parietal situation where the task does not use parietal processing cortex [7,33]. When processing of the relevant feature resources. Participants had to discriminate between a circle depended on parietal cortex, as is the case for orientation and a square. Shape discrimination has repeatedly been processing (exps. 1 and 4), there was an effect of the shown to rely on IT in the ventral stream [13,21,39]. digit s magnitude on response times. Conversely, there was Subjects made an average of 3.9% errors, with a no effect of the irrelevant digit on the processing of color maximum of 15.3% errors made by one subject. No (exps. 2 and 3) or shape (exp. 5), which relies only speed accuracy trade-off was present, as indicated by the minimally on parietal resources. absence of a negative correlation between RT and number Importantly, alternative accounts can be ruled out. of errors, computed over 18 data couples (nine numbers, Firstly, there was no dependency of the effect on elapsed separated for left and right responses), r520.06; n518; processing times. In exps. 3 and 5 conditions were created P,0.40. Mean RTs of correct responses with the digits 1 such that no SNARC effect was observed with similar or to 9 were, respectively, 447, 458, 461, 462, 464, 466, 449, even longer response times than those observed in con- 459 and 495 ms. The regression analysis of repeated ditions which elicited a reliable SNARC effect (exp. 1). measures data resulted in the following equation (shown in Secondly, in contrast with Ref. [22], mechanisms involved Fig. 5): in object-based attention cannot account for the pattern of semantic influences in our experiments, with number as drt magnitude. irrelevant information. Indeed, the SNARC effect did not depend on whether the relevant information consisted of a No SNARC effect was found as there was no significant 2D object. With a 2D object as a target against the contribution of number magnitude to the pattern of left numerical background a SNARC effect was observed in an hand right hand differences (t(19)520.35, S.D.54.5, orientation identification task, but not in a shape discrimi- P,0.3; six participants with a negative slope). This shows nation task. Importantly, the 2D objects were of equal that the unattended digits were not processed up to the complexity in both tasks. Thirdly, there was no difference level of semantic information. in dimensional overlap between the experiments [24,48]:

7 W. Fias et al. / Cognitive Brain Research 12 (2001) Fig. 5. Differences in RT (drt) between right and left hand responses (right left) as a function of the irrelevant digit in shape discrimination (circle or square). Squares indicate the observed drts. The continuous line depicts the predicted drts on the basis of the regression analysis. neither the processing of color, shape or orientation kept the response dimension constant in all our experiinvolves stimulus or response representations that are ments, it becomes clear that the SNARC effects must have related to the numerical value represented by a digit. emerged at a stage of information processing before motor Having excluded alternative explanations, we can con- control, that is, while encoding perceptual/ cognitive repreclude that the efficiency of feature-based attention is sentations. Representations of line orientation would comdetermined by the neural structures that are involved in the pete with representations of number magnitude in parietal processing of both relevant and irrelevant information. If cortex. Consequently, the SNARC effect would already be the processing of target and distracter make use of (partly) implied in the neural code that prefrontal cortex receives shared neural circuits then feature-based attention is less from parietal cortex to compute the appropriate response. efficient compared to the situation where the processing of This leads us to speculate that stronger SNARC effects target and distracter is mediated by distinct circuits. These might be obtained if the response programming itself also observations are in line with the view that feature-based takes place in parietal cortex, in a visually guided motor attention modulates processing in feature-specific cortical task. Before future research tests this hypothesis, however, areas [6,8,9,29,34,41]. Our results show that this relative we can already conclude from the present data that neural enhancement is less efficient when there is neural overlap overlap between target and distracter processing in parietal between target and distracter processing. Whether relative cortex contaminates response times even in tasks with enhancement is accomplished by facilitation or by inhibi- arbitrary stimulus response association rules. tion cannot be distinguished with the present paradigm. With neural overlap we imply that the signal-to-noise Our current task, aimed at studying feature-based atten- ratio of the information coded by a neural unit, be it a tion, naturally also implies a motor component: particip- single neuron or a cell assembly, can be directly affected ants had to associate the visual information with either a by irrelevant afferents. For instance, the tuning curve of a left-hand or right-hand response. This type of task, then, neuron to a relevant feature (e.g. upward orientation) may involves learning an arbitrary rule to convert visual be affected flattened or even sharpened depending information into a motor command, as opposed to the on whether task-irrelevant input to the neuron coincides more straightforward rules that underlie visually guided with the relevant input. Such influences would not occur if motor processing. Passingham et al. [40] have suggested the irrelevant information uses a different neural pathway. that the former type of processing is regulated primarily by This hypothesis borrows the concept of overlap from the prefrontal cortex, whereas parietal cortex would be more dimensional-overlap theory [24,48], but suggests that involved in visually guided motor processing [36]. Thus, it similarity of neural circuits, rather than similarity of may seem surprising at first that we find interference on stimulus and/ or response, is the basis of interference. In parietal processing at all in our task with arbitrary visuo- this sense, it is an extension of the dimensional-overlap motor associations. However, considering the fact that we theory: it predicts interference in all of the cases of

8 422 W. Fias et al. / Cognitive Brain Research 12 (2001) dimensional overlap, but also in cases where dissimilar [8] M. Corbetta, F.M. Miezin, S. Dobmeyer, G.L. Shulman, S.E. Petersen, Attentional modulation of neural processing of shape, stimuli and/or responses are processed (at least in part) by color, and velocity in humans, Science 248 (1990) the same neural structures. [9] M. Corbetta, F.M. Miezin, S. Dobmeyer, G.L. Shulman, S.E. Obviously, the present evidence from behavioral experi- Petersen, Selective and divided attention during visual discriminations ments can only indirectly support a hypothesis of neural of shape, color, and speed: functional anatomy by positron overlap. We depend on the existing literature to identify emission tomography, J. Neurosci. 11 (1991) [10] S. Dehaene, R. Akhavein, Attention, automaticity, and levels of neural circuits involved in discriminating color, shape or representation in number processing, J. Exp. Psychol.: Learn. orientation, and in autonomous number processing. Event- Memory Cogn. 21 (1995) related functional imaging and/ or single-unit study using [11] S. Dehaene, S. Bossini, P. Giraux, The mental representation of the present or a similar paradigm is required to provide parity and number magnitude, J. Exp. Psychol.: Gen. 122 (1993) direct evidence for our neural-overlap hypothesis. Some [12] S. Dehaene, L. Cohen, Towards an anatomical and functional model preliminary evidence, however, does already exist with of number processing, Math. Cogn. 1 (1995) primate prefrontal neurons in visual discrimination tasks [13] R. Desimone, T.D. Albright, C.G. Gross, C. Bruce, Stimulusselective properties of inferior temporal neurons in the macaque, J. [3,27]. For instance, Bichot and Schall showed that Neurosci. 4 (1984) neurons in the frontal eye field coding the saccade target [14] J. Duncan, Selective attention and the organization of visual based on one visual feature (e.g. a red color) responded information, J. Exp. Psychol.: Gen. 113 (1984) more clearly if the irrelevant feature (e.g. a square shape) [15] M.J. Eacott, D. Gaffan, The role of monkey inferior parietal cortex indicated the previous saccade target than if it did not. In in visual-discrimination of identity and orientation of shapes, Behav. other words, an irrelevant stimulus response association Brain Res. 46 (1991) [16] I. Faillenot, J. Decety, M. Jeannerod, Human brain activity related to directly affects the signal-to-noise ratio of neurons whose the perception of spatial features of objects, Neuroimage 10 (1999) code pertains to the same response dimension. Analogous processes of neural overlap in parietal cortex may account [17] I. Faillenot, S. Sunaert, P. Van Hecke, G.A. Orban, Orientation for our present data, with semantic influences from the discrimination of objects and gratings compared: an fmri study, Eur. J. Neurosci. 13 (2001) number on feature processing in the dorsal pathway, but [18] W. Fias, M. Brysbaert, F. Geypens, G. d Ydewalle, The importance not in the ventral pathway. of magnitude information in numerical processing: evidence from the SNARC effect, Math. Cogn. 2 (1996) [19] W. Fias, P. Dupont, B. Reynvoet, S. Raiguel, L. Mortelmans, G.A. Orban, A dissociation between ventral and dorsal visual streams Acknowledgements reflecting the nature of the cognitive task, Soc. Neurosci. Abs. 26 (2000) Thanks are due to Marc Brysbaert, Mauro Pesenti and [20] L. Girelli, D. Lucangeli, B. Butterworth, The development of two anonymous reviewers for thoughtful comments. This automaticity in accessing number magnitude, J. Exp. Child Psychol. research is supported by grant D of the Flemish 76 (2000) [21] C.G. Gross, P.H. Schiller, C. Wells, G.L. Gerstein, Single unit Fund for Scientific Research. Johan Lauwereyns is supactivity in temporal association cortex of the monkey, J. Neuroported by grant RFTF96L00204 from the Japanese Society physiol. 30 (1972) for the Promotion of Science. [22] G.W. Humphreys, M. Boucart, Selection by color and form in vision, J. Exp. Psychol.: Hum. Percept. Perform. 23 (1997) [23] J. Jonides, Voluntary versus automatic control of the mind s eye s References movement, in: J.B. Long, A.D. Baddeley (Eds.), Attention and Performance, Vol. IX, Erlbaum, Hillsdale, NJ, 1981, pp [24] S. Kornblum, T. Hasbroucq, A. Osman, Dimensional overlap: [1] Y.G. Abdullaev, K.V. Melnichuk, Counting and arithmetic functions cognitive basis for stimulus response compatibility a model and of neurons in the human parietal cortex, NeuroImage 3 (1996) 216, a taxonomy, Psychol. Rev. 97 (1990) Abstract. [25] D. LaBerge, Attentional Processing, Harvard University Press, [2] T. Allison, G. Mccarthy, A. Nobre, A. Puce, A. Belger, Human Cambridge, MA, extrastriate visual cortex and the perception of faces, words, [26] J. Lauwereyns, M. Koizumi, M. Sakagami, O. Hikosaka, S. numbers, and colors, Cereb. Cortex 4 (1994) Kobayashi, K. Tsutsui, Interference from irrelevant features on [3] N.P. Bichot, J.D. Schall, Effects of similarity and history on neural visual discrimination by macaques (Macaca fuscata): a behavioral mechanisms of visual selection, Nat. Neurosci. 2 (1999) analogue of the human Stroop effect, J. Exp. Psychol. Anim. Behav. [4] M. Brysbaert, Arabic number reading: on the nature of the numeri- Process. 26 (2000) cal scale and the origin of phonological recoding, J. Exp. Psychol.: [27] J. Lauwereyns, M. Sakagami, K. Tsutsui, S. Kobayashi, M. Gen. 124 (1995) Koizumi, O. Hikosaka, Responses to task-irrelevant visual features [5] L.L. Chao, A. Martin, Cortical regions associated with perceiving, by primate prefrontal neurons, J. Neurophysiol., in press. naming and knowing about colors, J. Cogn. Neurosci. 11 (1999) [28] R.F. Lorch, J.L. Myers, Regression-analyses of repeated measures data in cognitive research, J. Exp. Psychol.: Learn. Memory Cogn. [6] D. Chawla, G. Rees, K.J. Friston, The physiological basis of 16 (1990) attentional modulation in extrastriate visual areas, Nat. Neurosci. 2 [29] C.J. McAdams, J.H. Maunsell, Attention to both space and feature (1999) modulates neuronal responses in macaque area V4, J. Neurophysiol. [7] F. Chochon, L. Cohen, P.F. Van De Moortele, S. Dehaene, Differen- 83 (2000) tial contributions of the left and right inferior parietal lobules to [30] A.D. Milner, M.A. Goodale, The Visual Brain in Action, Oxford number processing, J. Cogn. Neurosci. 11 (1999) University Press, Oxford, 1995.

9 W. Fias et al. / Cognitive Brain Research 12 (2001) [31] B.C. Motter, Neural correlates of attentive selection for color or [39] K. Tanaka, Inferotemporal cortex and object vision, Annu. Rev. luminance in extrastriate area V4, J. Neurosci. 14 (1994) 2178 Neurosci. 19 (1996) [40] I. Toni, R.E. Passingham, Prefrontal-basal ganglia pathways are [32] A. Murata, V. Gallese, G. Luppino, M. Kaseda, H. Sakata, Selectivi- involved in the learning of arbitrary visuomotor associations: a Pet ty for the shape, size, and orientation of objects for grasping in study, Exp. Brain Res. 127 (1999) neurons of monkey parietal area AIP, J. Neurophysiol. 83 (2000) [41] S. Treue, J.C.M. Trujillo, Feature-based attention influences motion processing gain in macaque visual cortex, Nature 399 (1999) 575 [33] M. Pesenti, M. Thioux, X. Seron, A. De Volder, Neuroanatomical 579. substrates of arabic number processing, numerical comparison, and [42] J. Tzelgov, J. Meyer, A. Henik, Automatic and intentional processsimple addition: a PET study, J. Cogn. Neurosci. 12 (2000) 461 ing of numerical information, J. Exp. Psychol.: Learn. Memory 479. Cogn. 18 (1992) [34] M.A. Pinsk, S. Kastner, R. Desimone, L.G. Ungerleider, Expectation [43] A.H.C. Van der Heijden, Selective Attention in Vision, Routledge of color or face stimuli differentially modulates activity in human and Kegan Paul, London, UK, visual cortex, Soc. Neurosci. Abs. 26 (2000) 952. [44] S.P. Vecera, M.J. Farah, Does visual attention select objects or [35] M.I. Posner, Orienting of attention, Q. J. Exp. Psychol. 32 (1980) locations?, J. Exp. Psychol.: Gen. 123 (1994) [45] V. Walsh, S.R. Butler, The effects of visual cortex lesions on the [36] M.F.S. Rushworth, P.D. Nixon, R.E. Passingham, Parietal cortex and perception of rotated shapes, Behav. Brain Res. 76 (1996) movement. 1. Movement selection and reaching, Exp. Brain Res. [46] S. Yantis, Multielement visual tracking: attention and perceptual 117 (1997) organization, Cogn. Psychol. 24 (1992) [37] E. Shikata, F. Hamzei, V. Glauche, R. Knab, C. Dettmers, C. Weiller, [47] S. Zeki, in: AVision of the Brain, Blackwell Science, London, 1993, C. Buchel, Surface orientation discrimination activates caudal and p anterior intraparietal sulcus in humans: an event-related fmri study, [48] H. Zhang, J. Zhang, S. Kornblum, A parallel distributed processing J. Neurophysiol. 85 (2001) model of stimulus stimulus and stimulus response compatibility, [38] M. Taira, R. Kawashima, K. Inoue, H. Fukuda, A PET study of axis Cogn. Psychol. 38 (1999) orientation discrimination, Neuroreport 9 (1998)