حواس پرتی و انتخاب هدف در مغز: مطالعه fMRI
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|38744||2010||8 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Neuropsychologia, Volume 48, Issue 11, September 2010, Pages 3335–3342
Abstract To attend successfully, a specification of what is currently relevant is necessary, but not sufficient. Irrelevant stimuli that are also present in the environment must be recognized as such and filtered out at the same time. Using functional magnetic resonance imaging, we showed that posterior brain regions in parietal, occipital and temporal cortex are recruited in order to ignore distracting visual stimuli, while the specification and selection of relevant stimuli is associated with differential activity in frontal cortex and hippocampal areas instead. The results thus suggest that the selection of relevant objects can be anatomically dissociated from the handling of competing irrelevant objects. The dissociation between the increased involvement of parietal and occipital cortex in handling distraction on one hand, and that of frontal cortex in target specification on the other provides neurophysiological support for models of attention that make this functional distinction.
نتیجه گیری انگلیسی
2. Results and discussion Analysis of variance showed that participants responded more accurately in the conditions that did not require spatial filtering of any kind (92.7%) than in the ones that did (86.7%), F(1, 16) = 14.76, MSE = 0.004, p < 0.001, and similarly more so in the salience-based selection conditions (94.3%) than in the feature-based ones (85.1%), F(1, 16) = 7.24, MSE = 0.020, p < 0.05. Performance was lowest when participants had to both filter and perform feature-based selection (80.7%), F(1, 16) = 4.54, MSE = 0.003, p < 0.05. Reaction times mirrored these results perfectly. Participants were faster when they did not need to filter (671 ms) compared to when they did (724 ms), F(1, 16) = 37.43, MSE = 1266.039, p < 0.001, and were faster when using salience-based selection (662 ms) rather than when they were using feature-based selection (732 ms), F(1, 16) = 31.32, MSE = 2643.245, p < 0.001. The longest reaction times were recorded when participants had to both filter and perform feature-based selection (772 ms), F(1, 16) = 12.45, MSE = 987.623, p < 0.005. The behavioral results are shown in Fig. 2. Behavioral results of the experiment. On the left, bar graphs show mean accuracy ... Fig. 2. Behavioral results of the experiment. On the left, bar graphs show mean accuracy in percent correct. On the right, reaction times are plotted. Error bars represent one standard error of the mean. Figure options In the fMRI data, the comparison of the two feature-based search conditions with the salience-based ones showed that feature-based target selection resulted in increased activity in the right hemisphere in the parietal lobe (t(16) = 10.32, p < 0.001, MNI 18, −60, 56; extending to 24, −66, 58), and the middle temporal gyrus (MTG; t(16) = 8.77, p < 0.01, MNI 34, −80, 16), compared to salience-based selection 1. There was also increased activity associated with the salience-based selection conditions, which took the shape of a rather distributed set of clusters involving hippocampal areas on one end (t(16) = 12.41, p < 0.001, MNI −24, −34, −8 and t(16) = 9.31, p < 0.01, MNI 36, −16, −18), and the medial frontal cortex on the other (t(16) = 9.07, p < 0.01, MNI 0, 60, 22). In-between, the anterior cingulate was also involved in this contrast (t(16) = 11.08, p < 0.001, MNI 6, 38, 4). The individual contrasts revealed a pattern of results consistent with the above. In the first contrast, when participants had to filter simultaneously presented salient distractors while performing feature-based selection, compared to the corresponding condition in which distractors were not simultaneously presented (this contrast is labeled as “1” in Fig. 1), regions in bilateral temporal and occipital cortex were more activated (Fig. 3 and Table 1). Activity peaked around the MTG (t(16) = 9.92, p < 0.005, MNI 42, −76, 16 and t(16) = 8.75, p < 0.01, MNI 38, −78, 8), and the middle occipital gyrus (MOG; t(16) = 9.37, p < 0.01, MNI −36, −86, 8 and t(16) = 9.16, p < 0.01, MNI −46, −78, 4). t-Maps of the contrast between the feature-based selection & spatial filtering ... Fig. 3. t-Maps of the contrast between the feature-based selection & spatial filtering condition and the feature-based selection & no filtering condition, superimposed on a 3D rendering of a standard brain. In this and subsequent figures, the top left image shows a posterior view, the bottom left image shows an anterior view, and the image on the right a dorsal view. Label abbreviations are explained in the main text. Figure options Table 1. Regions of differential activity implicated in the individual contrast (see main text), with MNI coordinates and Z-scores. Feature-based selection: filtering > no filtering Region x y z Z-score Middle temporal gyrus 42 −76 16 5.54 38 −78 8 5.23 Middle occipital gyrus −36 −86 8 5.40 −46 −78 4 5.34 Table options The same comparison while performing salience-based selection (i.e., when attending to any orientation) between simultaneous spatial filtering and its counterpart without filtering showed no reliable differences (contrast 2 in Fig. 1). Note that the physical appearance of these displays was identical to those compared previously in the feature-based selection conditions. In other words, the increased temporal and occipital activity observed in the simultaneous filtering condition compared to the no filtering condition during feature-based selection was elicited by the specific need to filter out simultaneous distractors, and not by aspects of the physical appearance of the stimuli. Thus, the increased posterior activity observed in the feature-based selection condition (contrast 1 in Fig. 1) may be taken as converging evidence for the role of occipital cortex and adjacent regions as an attentional filter, a notion broadly compatible with previous studies (e.g., Wojciulik & Kanwisher, 1999). In the third contrast between feature- and salience-based selection in the absence of spatial filtering (see Fig. 1), activity peaked in the right hemisphere in the MTG (t(16) = 10.79, p < 0.001, MNI 60, −2, −14). This locus was more anterior and relatively far away from the MTG clusters implicated in the previous contrast, however. In addition, clusters of activity showed up in hippocampal areas (t(16) = 9.51, p < 0.005, MNI 38, −22, −14 and t(16) = 9.28, p < 0.01, 32, −12, −18). As shown in Fig. 4 (and Table 2), at the same time no reliable differences were significant in parietal and occipital areas, suggesting that distraction played less of a role here. The increased hippocampal activation in the salience-based selection condition may be explained as follows. It may have been a result of all salient stimuli qualifying as targets in those conditions, as compared to the subset of stimuli in the feature-based selection conditions. It may thus be that more resources related to working memory would be recruited during salience-based selection as well, simply because more stimuli had to be encoded for response selection (cf., Linden et al., 2003). Crucially, this effect was not associated with spatial filtering. In other words, these mechanisms were apparent even when salient distractors were not shown simultaneously with targets. t-Maps of the contrast between feature-based selection & no spatial filtering ... Fig. 4. t-Maps of the contrast between feature-based selection & no spatial filtering and salience-based selection & no filtering, superimposed on a 3D rendering of a standard brain. Figure options Table 2. Regions of differential activity implicated in the individual contrast (see main text), with MNI coordinates and Z-scores. No filtering: salience-based > feature-based selection Region x y z Z-score Middle temporal gyrus 60 −2 −14 5.74 Hippocampal 38 −22 −14 5.43 32 −12 −18 5.37 Table options We finally compared directly between feature-based selection and salience-based selection in the spatial filtering conditions (contrast 4 in Fig. 1). According to our hypothesis, this contrast should reveal both more activity in posterior regions due to the need to perform spatial filtering (observed during feature-based selection), as well as more activity in frontal regions for salience-based selection. This was indeed confirmed, as can be seen from Fig. 5. First, we observed increased activity in the feature-based selection condition, which closely resembled the ones reported previously (in the main effect analysis and the individual contrast shown in Fig. 3). Activity peaked in the MOG (t(16) = 9.94, p < 0.005, MNI 38, −78, 12), in superior parietal regions (t(16) = 9.35, p < 0.01, MNI 18, −68, 64 and t(16) = 8.79, p < 0.01, MNI 12, −70, 58), and bilaterally in the precuneus (t(16) = 9.60, p < 0.005, MNI −20, −68, 52 and t(16) = 8.97, p < 0.01, MNI 32, −68, 36). This pattern of parietal and occipital activity resembled the one in the first individual contrast, which supported the idea that these areas are specifically involved in the process of filtering simultaneous distractors. Second, salience-based selection resulted in increased frontal activity compared to when participants were performing feature-based selection, peaking around the medial frontal gyrus (MFG; t(16) = 13.17, p < 0.001, MNI 2, 56, 24 and t(16) = 9.51, p < 0.005, MNI 4, 64, 20), and extending towards its superior part (t(16) = 9.42, p < 0.005, MNI 2, 58, 8). These results were similar to those reported in the ‘main effect’ contrast between feature- and salience-based search. The need to change target selection criteria, or to look for a specific target template match as compared to using an unspecific, salience-based search mode thus affected distinctly different brain regions than those involved in dealing with the presence of salient distractors. Taken together, the functions of both anterior and posterior regions contributed to the contrast between feature- and salience-based selection during spatial filtering. Table 3 lists the clusters of differential activity that were implicated in the analyses presented above. t-Maps of the contrast between feature-based selection & spatial filtering and ... Fig. 5. t-Maps of the contrast between feature-based selection & spatial filtering and salience-based selection & spatial (stimulus-level) filtering, superimposed on a 3D rendering of a standard brain. SP = superior parietal lobe. Figure options Table 3. Regions of differential activity implicated in the individual contrast (see main text), with MNI coordinates and Z-scores. Filtering: feature-based > salience-based selection Filtering: salience-based > feature-based selection Region x y z Z-score x y z Z-score Middle occipital gyrus 38 −78 12 5.54 Precuneus −20 −68 52 5.46 32 −68 36 5.29 Superior parietal lobe 18 −68 64 5.46 12 −70 58 5.24 Medial frontal gyrus 2 56 24 6.21 4 64 20 5.44 2 58 8 5.41 Table options In conclusion, the present study produced two principal findings. First, when the search task required filtering of (salient) distractors, areas across parietal, temporal, and occipital cortex became (more) involved. Second, when target selection did not require specific matching of a target template (i.e., when any salient stimulus was a target), the differentially involved brain regions were mostly in frontal cortex and hippocampal areas. When both spatial filtering of distractors and feature-based target selection were needed, then both areas were implicated. Our results should not be taken to mean that activity in parietal and occipital cortex are not at all modulated by processes related to target selection or its associated difficulty in other tasks, as shown by others (Donner et al., 2003, Martínez et al., 1999 and Reynolds and Desimone, 2003), since many factors related to this were explicitly held constant in our study. Indeed, attentional filtering in the inferior parietal sulcus (IPS) is most likely accomplished by boosting target-related visual signals, in contrast to local mutual suppression observed in extrastriate cortex (Kastner, De Weerd, Desimone, & Ungerleider, 1998), as activation in this region has been associated with both perceptual visibility and perceptual interference (Marois, Chun, & Gore, 2003). However, in the present study, given the absence of a comparable difference in these posterior brain areas between feature- and salience-based selection in the absence of spatial filtering, as well as the observed increase in frontal activity observed in both of the salience-based selection conditions, our hypothesis that changing target selection criteria does not directly modulate activity in posterior regions was supported. The experimental division between target selection and distractor filtering was mirrored by a relatively clear anatomical and functional organization of attention, and can explain recent reports of distinct causal influences of frontal and parietal cortex on activity in the visual cortex (Ruff et al., 2008). The results support the idea that frontal cortex deals with the specification, consolidation, and selection of target stimuli, while parietal cortex deals with filtering those that are not targets. A consequence implicit in this dissociation is that filtering of simultaneously presented distractors is not necessary when selection is salience-based; and indeed did not elicit increased activity in parietal regions. These became more strongly involved only when a feature-based target selection had to be made amongst salient distractors. The increased frontal activity appeared whenever the selection of any salient stimulus was contrasted with the selection of a specific kind, regardless of the composition of the display. One caveat with the interpretation of the results is that the behavioral difficulty of the conditions was also different. Thus, some of the activity observed presently may be attributable to ‘difficulty effects’, and hence potentially implicate the involvement of the so-called “default network” (Shulman et al., 1997). The degree to which this could be the case cannot be easily assessed, but even if difficulty causes brain activity on its own, the cause of this difficulty in the present paradigm still lies in the differences between selection and filtering requirements. Taken together, the observed degree of functional specificity of the brain supports the framework of attention originally proposed by Duncan (1980), Desimone and Duncan (1995), Duncan and Humphreys (1989) in that the functions defined as “target selection” and “distractor filtering” were found to modulate distinct cortical sites. This theory is also in line with recent findings of occipital activity elicited by both the location of an expected target as well as that of a distractor, which is more problematic to explain for accounts that focus on the distinction between top-down and bottom-up processing (Brefczynski and DeYoe, 1999 and Ruff and Driver, 2006). Thus, the present results suggest that the co-operation of selection and filtering processes provides a comprehensive account of two main functions of attention.