Does the GNAT incongruent trial make cognitive conflict?
Brain activities during Go/No-Go Association Task
Koichi Hioki ,
2 2 2 2
Seiji Nakagawa , Hiroshi Watanabe , Hiroyuki Umemura , Katsunori Matsuoka
& Minoru Karasawa
Graduate School of Humanities and Social Science, Kobe University, Japan,
Institute for Human Science and Biomedical Engineering, National Institute of Advanced Industrial Science and Technology (AIST), Japan, Kobe
INTRODUCTION RESEARCH INTEREST
The aim of this study was to examine brain activities with We argue that GNAT is a tool which measures the strength of the
magnetoencephalography (MEG) during the Go/No-Go Association implicit association between a concept and an attribute category. If this
Task (GNAT). is true, while discriminating “Bugs” and “Good” from distracters(
incongruent trials), we should be inducing a conflict state in our
participants. On the other hand, while discriminating “Bugs” and
Social psychologists typically use questionnaires to study attitude
“Bad” from distracters( congruent trials), a conflict state will not occur.
toward specific objects (e.g., a minority group, self ). For example
participants are asked, “Do you think that African Americans We believe that this could be tested and verified neurophysiologcally.
demand equal rights too much?” or “ Do you think that media and
the government pay respect too much to African Americans?” Reaction latency “Bugs or Bad” trials
However, as Wittenbrink, Judd, & Park (1997) pointed out, biases － ) Reaction latency “Bugs or Good” trials
and distortion cannot be completely prevented when using self- GNAT score of “Bugs”
report questionnaires. ↓
In response to such problems, Greenwald, McGhee, & Schwartz We assume it reflects implicit positive impression towards Bugs.
(1998) created another measurement of social stereotypes. They However, there are an unsettled question of whether the GNAT score truly
noted that there are differences in reaction time to stereotype words reflects “Implicit Association".
between high and low prejudice person. They created the Implicit
Association Test (IAT) as a measure for implicit prejudice. Later,
Nosek ＆ Banaji (2001) created the Go/No-Go Association Task Bug and Happy？
(GNAT) as an improved version of the IAT. They assumed that a
high prejudice person has greater association between African
? ←This can be tested neurophysiologcally
Americans and negative words than a low prejudice person. For that
reason, they predicted that a high prejudice person would be faster
than a low prejudice person in identifying the pair words “Bad” and vWhile performing the incongruent trials,
“Black“ among other stimuli. While in the case of identifying a cognitive conflict state should be induced.
“Black” and “Good,” a low prejudice person would be faster than a vAnd while performing the congruent trials (“Bugs” and
“Bad”), a conflict state will not be induced.
high prejudice person.
Table 1: Setting of the GNAT METHOD
Participants: 5 Japanese graduate and undergraduate students.
High prejudice Low prejudice Trials: Participants were provided informed consent and sat in MEG
(Neuromag-122 ,4-D Neuroimaging Ltd., Finland) which settled in a
Black + Bad Easy Difficult shielded room. Participants then performed two types of block. One was
the Congruent Block (identifying “Bugs” and “Bad” from distracters) and
the other was the Incongruent Block (identifying “Bugs” and “Good” from
Black + Good Difficult Easy
distracters). Both blocks contain 200 trials, one half of which was deemed
to elicit “Yes” responses (press a button) while the other half designed for
“No” responses (ignore).
Data Sampling: During the GNAT task, participants were scanned by 122
The GNAT is a dual task which requires both (1) classification coils of MEG. Data was sampled by 400Hz and digitally filtered
judgments and (2) attribute judgments for stimulus items. The frequencies below 0.5Hz and above 30Hz.
GNAT consisted of two blocks, one of the blocks constructed
with “Good” as the target attribute, whereas “Bad” being the
target in the other half. In sum, the GNAT consist of “identifying
target category or Bad” trials (e.g. identifying “Bugs” and “Good”
from distracters) and “identifying target category or Good” trials
(e.g. identifying “Bugs” and “Bad” from distracters).
We calculate the GNAT score by subtracting reaction latency of “Bugs
or Good” trials from “Bugs or Bad” trials. Therefore, it is thought that
the GNAT score reflects the implicit positive impression one has
towards Bugs. Thus the GNAT is considered to be a tool which
measures the strength of the implicit association between a concept
and an attribute category.
GNAT score: As shown in Table 2, all participants showed shorter reaction A
latency in the Congruent trials (“Bugs or Bad” ) than in the Incongruent
trials (“Bugs or Good”) F (1, 4) > 64.31, p <.01. In other words, the
Congruent trials were easier to perform than the Incongruent trials.
Table 2: All participants’ reaction latencies and GNAT scores (ms)
MEG data: We calculated 12 root means of each hemisphere of 5 participants. P A:anterior
The coils used to calculate are represented as red triangles in Fig.1 (all MEG data P:posterior
shown in Appendix).
Fig.1: Graphs of root meant MEG data
Condition (2: Congruent / Incongruent) *Red line represent Incongruent trials and blue line represent Congruent
x Time (66: divided from -150ms to 1500ms into 66 time windows; trials.
25ms each) **Higher scores means higher activation of brain region
ANOVA on root meant MEG data.
*** 0ms indicate onset timing of stimulus
We found some significant interactions between Condition and Time at the right
hemisphere Fs (65,260) > 1.37, ps < .05. And follow–up analysis of Time
indicated simple main effects of Condition at about 300ms – 500 ms after
stimulus onset ( these results are indicated as green circles in Fig.1).
Analysis of reaction time showed that incongruent trials took longer than congruent trials. Furthermore, the MEG data
revealed brain activities over midline central scalp regions (above the Anterior Cingulate Cortex, ACC) in only
incongruent trials. These activities occurred at about 400 ms after the stimulus appeared (i.e., about 250 ms before the
response). In conclusion, our data suggest that incongruent GNAT trials induce cognitive conflict and that the time lag
is caused by cognitive conflict during stimulus judgment.
Furthermore, MEG coils have a useful characteristic that the scanned data reveal brain region activities which located
directly below the coil. Therefore, we can assume that the GNAT incongruent trials induced activities of the ACC and
the left inferior parietal lobule at about 300ms after stimulus (shown as red circles in Fig.2). And it is known that when
a conflict state occurs, the ACC will also be activated (Fan, Flombaum, McCandliss, Thomas, & Posner, 2003). Fig.2: MEG coils (Right Side View)
Therefore, the results suggest that the GNAT time-lag(score) may reflect cognitive conflict and such conflict is
induced before participants respond.
APPENDIX: All MEG data
vFan, Flombaum, McCandliss, Thomas, &
Posner, 2003, Cognitive and Brain
Consequences of Conflict, NeuroImage,18, 42
vGreenwald, McGhee, & Schwartz, 1998,
Measuring Individual Differences in Implicit
Cognition: The Implicit Association Test,
JPSP,74 (6), 1464-1480.
vNosek, ＆ Banaji, 2001, The Go/No-Go
Association Task, Social Cogn. 19 (6),625-
vWittenbrink, Judd, ＆ Park, 1997, Evidence
for Racial Prejudice at the Implicit Level and
Its Relationship With Questionnaire Measures,
MEG data of left hemisphere MEG data of right hemisphere JPSP,72 (2), 262-274.
Whole-head neuromagnetometor (Nuromag-122 , Nuromag Ltd., Finland)