Twenty-four right-handed subjects, 13 males and 11 females, participated in the experiment. Handedness was assessed by a revised version of Annett's hand preference questionnaire.^13,14 Exclusion criteria included a current or past history of psychiatric disorders, traumatic brain injury, neurological illness, alcohol or drug abuse, or relevant visual defects.

All subjects fulfilled the inclusion criteria of age 18 to 35 and more than 12 years of education. All subjects gave informed written consent after the nature of the experiment was explained.

The experiment used a blocked paradigm and consisted of the attitude attribution task (experimental condition) and gender matching task (a control condition), and baseline crosshair.

To understand a person's inner attitude toward another person, the successful recognition and evaluation of facial affect is essential. However, because even the same facial affects have different meanings according to context, we should attend to and use contextual information. For example, if we observed a man who was happy to see the failure of his colleague, we would understand him as having a malevolent attitude toward his colleague. On the contrary, if we noticed someone else's happy facial affect at the success of his or her colleague, we would understand him or her as having benevolent attitude toward that person. Therefore, to examine the ability to attribute other's inner attitude correctly by observing facial affect in an emotional context, we made stimuli that are composed of two pictures. One was a picture depicting facial affect, and the other was a picture displaying an affectively-loaded situation. The pictures depicting various facial affects were selected from color photographic sets of faces that were standardized by Gur et al.¹⁵ The pictures displaying an emotionally-laden situation were selected from the International Affective Picture System.^16,17

The preliminary stimuli made by the above method were presented to groups of raters (n=32) who were asked to judge whether the protagonist in the picture depicting facial affect presented in one side had a benevolent or malevolent inner attitude toward the person displayed in the picture of the other side which provided contextual information.

To prevent confusion, prior to this process, the raters were instructed that the facial affect in the former picture was expressed in the situation where the person displayed in the contextual picture was not able to notice it, and that only the rater was able to notice the facial affective expression in the third person perspective.

For the purpose of constructing a valid task, a subset of stimuli was selected based on the high degree (>90%) of concordance by the raters. This final subset, administered during the fMRI study, comprised a total of 20 pictures. In five stimuli, the person with pleasant facial affect presented on the right side had a benevolent inner attitude toward the person in the picture presented on the left side. These five stimuli represent the typical situation in which a person is happy to see another's success, happiness, or joy. In another five stimuli, though the person in the picture presented on the right side showed a pleasant facial affect, he or she had a malevolent attitude toward the person in the picture presented on the left side. These five stimuli represent the situation in which a person is happy to see another's disaster or misery. In another five stimuli, in spite of an unpleasant facial affect, the person presented on the right side has a benevolent attitude toward the person in the picture presented on the left side. These five stimuli typically represent the situation in which a person is sad to see another's disaster or misery. In five additional stimuli, the person with the unpleasant facial affect presented on the right side has a malevolent attitude to the person in the picture presented on the left side. These five stimuli typically describe the situation in which a person expresses disgusting facial affect toward another's behavior. The mean concordance rate of this final subset was 97.36±2.91%.

During the scanning, the subjects were asked to perform the same task described above, and to respond by pressing either the left button (benevolent attitude) or right button (malevolent attitude). In the gender matching task, the subjects were asked to judge whether the gender of people displayed in the two pictures was matched (left button press) or not (right button press). All subjects used right hand for responding.

The ratio (1 : 1) of benevolent versus malevolent attitude in the experimental condition was equal to the ratio of matched versus unmatched gender in the control condition to maintain the same proportion of motor responses. Facial affective pictures of the same individual were not repeated within one block. Accuracy and reaction times data were collected during the scanning session. A practice session with different stimuli was conducted prior to initiating the scan to ensure optimal performance.

In both the experimental task and control task, each stimulus was presented for 4.1 seconds, followed by a 1-second interstimulus interval. Each task block contained 20 trials of either experimental task or control task, resulting in a total of 102-second blocks. Thirty-second blocks of baseline crosshair followed each block, and each scanning session comprised four blocks (A-R-C-R or C-R-A-R, where A indicates activation condition and C indicates control condition and R indicates baseline crosshair). During the baseline crosshair block, subjects gazed at a cross (+) on the screen. Each subject underwent two scanning sessions. The presentation order was counterbalanced across subjects (A-R-C-R-C-R-A-R or C-R-A-R-A-R-C-R).

MRI imaging was performed with a 3.0 T imager (ISOL technology, Korea). Before imaging, a global shimming procedure, using first- and second-order shims, was performed to optimize the magnetic field over the imaging volume of interest.

Whole-brain axial T1-weighted anatomical images were acquired using an inversion recovery sequence (4 mm contiguous slices; TE, 16 ms; TR, 3,200 ms; matrix size, 256×192; FOV, 22×22 cm). During the study, echo planar functional images were acquired using a gradient echo pulse sequence (4 mm contiguous slices; TE, 35 ms; TR, 3,000 ms; matrix size, 64×64; FOV, 22×22 cm).

All image data were analyzed with Statistical Parametric Map software (SPM99, Wellcome Department of Cognitive Neurology, London, UK).

By using the first image in the time series as the reference image, movement correction was performed. The images were co-registered and resliced, and the sampling errors were adjusted (sinc interpolation). All images were normalized to Montreal Neurological Institute coordinates and smoothed with an isotropic Gaussian kernel (10 mm full-width at half-maximum). Analysis was carried out using the general linear model and a boxcar waveform convolved with a hemodynamic response function accounting for the delayed cerebral hemodynamic response after stimulus presentation. Subject-specific, low frequency drifts in signal changes were removed by a high pass filter and global signal changes were treated as a covariate of no interest. Specific effects for each voxel were tested by applying appropriate contrast to the parameter estimates for each condition, resulting in t statistic for every particular voxel. The resulting set of voxel values for each contrast constituted a statistical parametric map of the t statistics SPM{t}. A second-level group analysis was performed using a random-effects model.

Activation results are reported at p<0.005 (uncorrected) with an extent threshold of 10 contiguous voxels for a priori regions of interest: inferior frontal cortex (BA 44/45), inferior parietal lobule, superior temporal sulcus.

Next, we explored the activation of non-a priori areas. For non-a priori areas, we reported results that survive a statistical height threshold of p<0.05, corrected for multiple comparisons, with an extent threshold of 10 contiguous voxels. Because we have strong a priori hypothesis, we thought this two step approach could be justified.

Performance was measured by percent correct and reaction time, but because performance data for four participants, 2 males and 2 females, were unscorable due to technical failure, the analyses were done for performance data for only twenty participants. As expected, participants did not have difficulties performing the tasks. For the attitude attribution task, they averaged 95.0±3.9% correct (range 87.5-100) with reaction time averaging 1904.6±481.9 msec (range 1012.0 to 2919.0). For the gender matching task, they averaged 95.1±3.4% correct (range 87.5-100) with reaction time averaging 1650.8±264.0 msec (range 1172.0 to 2297.0). Paired t-test showed that the accuracies (t=1.15, p=0.27) were not significantly different between the two tasks. However, the reaction time was longer for the attitude attribution task than for gender matching task (t=2.89, p<0.01).

Table 1 shows the results of subtracting the control condition (gender matching task) from the experimental condition (attitude attribution task). First of all, we explored the activation of the a priori region of interest at a threshold p<0.005, and observed that the activations of the brain regions, such as the left inferior frontal cortex (BA 44/45), left superior temporal sulcus (BA 22), and left inferior parietal lobule (BA 40) were activated during the intention reading task, as compared to the gender matching task (Figure 1). Next, we explored the activation of other areas and observed the activation of left inferior frontal cortex (BA 47).