4.7 Results 145
4.7.3 fMRI analyses – Semantic access 151
In our initial brain imaging analyses we aimed to define a network of brain regions that supports the ability to access semantic knowledge about famous names. In a second step, we then aimed to examine the extent to which NRB responses engage that same network. To define the semantic access network, we used the conjunction of two contrasts from the name-recognition phase and the occupation phase, respectively (Nichols, Brett, Andersson, Wager, & Poline, 2005). For this purpose, data from both phases were combined into a single General Linear Model (GLM).
To define semantic access in the first phase of the experiment, we
compared brain activity for famous names given ‘identify’ responses, which were presumably associated with recall of a distinct piece of semantic information, with that associated with famous names given ‘unfamiliar’ responses. To define
semantic access in the second phase of the experiment, we focused specifically on the forced-choice occupation trials, as they explicitly required access to semantic knowledge about the presented famous names. We designated each forced-choice
occupation trial as either reflecting or not reflecting successful semantic access, and we contrasted brain activity for these two types of trials. Our inferences about the success of semantic access in trial assignments took into account both the objectively scored response accuracy of the forced-choice judgments as well as the associated confidence expressed. Specifically, we defined our semantic-access contrast using a comparison between accurate forced-choice trials that were associated with confidence levels two through four, and all (i.e., accurate and inaccurate) forced-choice trials associated with the lowest confidence level. Notably, accurate and inaccurate forced-choice trials associated with the lowest confidence level were assumed to reflect chance guessing given that our
behavioral analyses revealed that accuracy was not different from chance for such trials.
Our first random-effects GLM thus included 6 predictors of interest. Separate predictors were created to model four types of trials associated with the three types of recognition response: famous identify (Fam_Iden), famous NRB (Fam_Nrb), famous unfamiliar (Fam_Unf), and fictional unfamiliar (Fict_Unf). Separate predictors were also created to model the two types of forced-choice occupation responses (i.e., those with and without semantic access; Fc_Sem and Fc_Nosem, respectively), and a corresponding predictor for confidence judgments was also included but was not examined. Finally, a confound predictor was
included to model inaccurate forced-choice trials associated with two, three, and four confidence levels, as well as corresponding confidence trials. The confound predictor also included trials for fictional names associated with NRB and
Figure 4-3: Brain regions implicated in semantic access, and in NRB responses
Sagittal views displaying brain areas involved in (A) conjunction analysis for successful semantic access (p < 0.005 for
each contrast, effective p < 0.000025, minimum cluster size= 13 voxels) and (B) contrast of NRB responses minus
‘unfamiliar’ responses, for famous names only (p < 0.001, minimum cluster size= 11 voxels). Main effects for each
analysis superimposed on representative participant’s structural MRI within the range of t-values shown for each analysis separately. Fam_Iden - famous names given ‘identify’ response, Fam_Nrb -famous names given NRB response, Fam_Unf –famous names given ‘unfamiliar’ response, Fict_Unf – fictional names given ‘unfamiliar’ response, Fc_Sem –
occupation forced choice trials associated with semantic access (see main text for details), Fc_Nosem – occupation forced choice trials associated with no semantic access.
Figure 4-4: Areas of overlap between semantic access and NRB responses
Separate analyses from Figure 4-3 overlaid on anatomical images of a representative participant’s brain, demonstrating regions of overlap. Activation map in blue reflects conjunction analysis for successful semantic access and that in orange reflects the Fam_Nrb > Fam_Unf contrast. Arrows highlight regions of overlap in (A) left middle posterior temporal gyrus
and (B) an inferior aspect of left ventrolateral prefrontal cortex. Beta weights for these two regions of overlapare
presented for descriptive purposes, for all conditions of the name-recognition phase runs and the two occupation forced- choice conditions of the occupation-phase runs. Blue and orange lines indicate contrasts used to isolate brain regions in the two separate analyses. Images shown in radiological convention (left equals right). Error bars represent SEM.
Table 4-4: Brain areas isolated in semantic access conjunction [Fam_Iden > Fam_Unf + Fc_Sem > Fc_Nosem]
Brain Regions x y z t-value
Cerebellum R 26 -68 -36 4.96
Brain stem M 8 7 6 7.82
Anterior cingulate M -4 46 18 8.31
Precuneus M -4 -56 21 6.74
Middle temporal gyrus L -61 -17 -9 7.83 Angular gyrus L -43 -65 18 6.54 Anterior temporal lobe L -44 16 -23 4.00 Ventrolateral prefrontal cortex L -30 12 -9 3.40
Threshold set at p < 0.005, effective p < 0.000025, minimum cluster size = 13. L, left; R, right, M, midline. Coordinates are expressed in millimeters in the Talairach and Tournoux brain atlas: x, medial–lateral axis (negative, left); y, anterior–
posterior axis (negative, posterior); z, dorsal–ventral axis (negative, ventral). Fam_Iden - famous names given ‘identify’ response, Fam_Nrb -famous names given NRB response, Fam_Unf –famous names given ‘unfamiliar’ response, Fict_Unf – fictional names given ‘unfamiliar’ response, Fc_Sem –forced choice trials associated with semantic access (see main text for details), Fc_Nosem – forced choice trials associated with no semantic access.
Table 4-5: Brain areas isolated in contrast of NRB responses minus ‘unfamiliar’ responses [Fam_Nrb > Fam_Unf]
Brain Regions x y z t-value
Caudate L -14 1 14 7.37
R 8 5 12 9.90
Ventrolateral prefrontal gyrus L -40 20 11 9.02
R -31 22 4 7.30
Dorsolateral prefrontal gyrus L -41 13 27 10.31
R -22 37 33 6.50
Superior parietal lobe L -34 -65 33 5.53 Middle temporal gyrus L -52 -38 -3 7.37
Brain stem M -3 -24 -2 8.04
SMA / cingulate M 0 2 57 12.83
Cerebellum M 5 -49 -39 5.71
‘identify’ responses from the name-recognition stage.