Avoiding non-independence in fMRI data analysis: Leave one subject out

Esterman, Michael; Tamber-Rosenau, Benjamin J.; Chiu, Yu-Chin; Yantis, Steven

doi:10.1016/j.neuroimage.2009.10.092

Cited by 235 publications

(186 citation statements)

References 17 publications

Supporting

Mentioning

184

Contrasting

Order By: Relevance

“…To examine the time courses in more detail, we next tested the difference between HC and LC at each time point in each cluster. To avoid circularity, we used leave-one-subject-out (LOSO) cross-validation (Esterman et al 2010). To correct for multiple comparisons, we employed a threshold of p < .005, based on the 10 tests in each region (i.e., 0.05/10; a Bonferroni correction across all tests together would almost certainly be overly conservative given the dependency between tests).…”

Section: Fmri Resultsmentioning

confidence: 99%

“…To analyze activation on a timepoint-by-timepoint basis within clusters identified in the whole-brain ANOVA while avoiding circular voxel selection, we employed the leave-one-subject out (LOSO) method described by Esterman (2010). In this method, the voxels selected for a given ROI in a given subject are determined by the whole-brain suprathreshold voxels identified in a wholebrain analysis carried out with all the other subjects in the study.…”

Section: Glm Analysismentioning

confidence: 99%

See 1 more Smart Citation

Multiscale Integration of Contextual Information During a Naturalistic Task

et al. 2017

View full text Add to dashboard Cite

Everyday experience requires rapid and automatic integration of incoming stimuli with previously stored knowledge. Prior knowledge can help construct a general "situation model" of the event, as well as aid comprehension of an ongoing narrative. Using fMRI in healthy adult humans we investigated processing of videos whose locations and characters were always familiar but whose narratives were either a continuation or non-continuation of an earlier video (high context (HC) or low context (LC) respectively). Responses in parahippocampal gyrus and retrosplenial cortex were composed of an initial transient, locked to the video onsets, followed by a period of lower amplitude activation that was greater in the LC condition. This may reflect rapid processing of core components of situation models such as location and characters and more gradual incorporation of their narrative themes. By contrast, activity increases in left hemisphere middle temporal gyrus (MTG), angular gyrus and inferior frontal gyrus were maintained throughout the videos and were higher for HC versus LC videos. Further, activity in the left MTG peaked earlier in the HC condition. We suggest that these regions support representations of the specific inter-linked concepts necessary to comprehend an ongoing narrative, which are already established for the HC videos.

show abstract

Section: Fmri Resultsmentioning

confidence: 99%

Section: Glm Analysismentioning

confidence: 99%

Multiscale Integration of Contextual Information During a Naturalistic Task

et al. 2017

View full text Add to dashboard Cite

show abstract

“…To determine if amygdala connectivity patterns predicted behavioral performance, multiple linear regression analyses were completed using a leave‐one‐subject‐out (LOSO) approach (Esterman, Tamber‐Rosenau, Chiu, & Yantis, 2010), which maintains independence between the training and testing datasets. Here, separate group‐level functional connectivity difference analyses using the left and right amygdala ROIs as seed regions were rerun 66 times, each time leaving one participant out.…”

Section: Methodsmentioning

confidence: 99%

Interpersonal early‐life trauma alters amygdala connectivity and sustained attention performance

et al. 2017

Self Cite

View full text Add to dashboard Cite

IntroductionInterpersonal early life trauma (I‐ELT) is associated with a myriad of functional impairments in adulthood, increased risk of drug addiction, and neuropsychiatric disorders. While deficits in emotional regulation and amygdala functioning are well characterized, deficits in general cognitive functioning have also been documented. However, the neural underpinnings of cognitive dysfunction in adults with a history of I‐ELT and the potential relationship between amygdala‐based functional connectivity and behavioral performance are currently poorly understood. This study examined how I‐ELT affects the cognitive and neural mechanisms supporting sustained attention.MethodsA total of 66 Veterans (18 with and 48 without a history of I‐ELT) completed a nonemotional sustained attention task during functional MRI.ResultsThe individuals with I‐ELT showed significant impairments in sustained attention (i.e., higher error rates, greater response variability). This cohort exhibited increased amygdala functional connectivity with the prefrontal cortex and decreased functional connectivity with the parahippocampal gyrus when compared to those without I‐ELT. These connections were significantly correlated with individual differences in sustained attention performance. Notably, classification analyses revealed that the pattern of amygdala connectivity across the whole brain was able to classify I‐ELT status with 70% accuracy.ConclusionThese results provide evidence of a lasting negative impact for those with a history of I‐ELT on sustained attention ability. They also highlight a critical role for amygdala functioning in cognitive control and sustained attention for those with a history of I‐ELT, which may underlie the observed attention deficits in clinical assessments and cognitive tests involving both emotional and nonemotional stimuli.

show abstract

“…We therefore extracted the parameter estimates at a voxel level threshold of P < 0.001. The fact that the cluster was not significant in each GLM is expected and can be explained by the loss of one degree of freedom due to the left-out subject and thus the slightly reduced statistical power (Esterman et al, 2010). Fig.…”

Section: Fmri Resultsmentioning

confidence: 85%

“…To get an unbiased estimate of the underlying effect sizes for each significant cluster, we used a leave one subject out crossvalidation procedure (Esterman, Tamber-Rosenau, Chiu, & Yantis, 2010). We repeatedly recalculated the second level GLM, iteratively leaving out one of the subjects in each calculation.…”

Section: Fmri Data Analysismentioning

confidence: 99%

Aversive smell associations shape social judgment

Homan

Yuan

Brosch

et al. 2017

Neurobiology of Learning and Memory

View full text Add to dashboard Cite

a b s t r a c tOnce associating another person with an unpleasant smell, how do we perceive and judge this person from that moment on? Here, we used aversive olfactory conditioning followed by a social attribution task during functional magnetic resonance imaging to address this question. After conditioning, where one of two faces was repeatedly paired with an aversive smell, the participants reported negative affect when viewing the smell-conditioned but not the neutral face. When subsequently confronted with the smell-conditioned face (without any smell), the participants tended to judge both positive and negative behaviors as indicative of personality traits rather than related to the situation. This effect was predicted by the degree of the preceding olfactory evaluative conditioning. Whole brain analysis of stimulus by stage interaction indicated differential activation of the ventromedial prefrontal cortex and right angular gyrus to the conditioned versus the neutral person during the attribution phase only. These results suggest that negative smell associations do not simply induce a negative perception of the target person but rather bias the attribution style towards trait attributions. The fact that this bias was evident regardless of behavior valence suggests it may reflect enhanced psychological distance. Thus, the known observation of social rejection triggered by aversive smell may be driven by a shift in social attribution style.

show abstract

Avoiding non-independence in fMRI data analysis: Leave one subject out

Cited by 235 publications

References 17 publications

Multiscale Integration of Contextual Information During a Naturalistic Task

Multiscale Integration of Contextual Information During a Naturalistic Task

Interpersonal early‐life trauma alters amygdala connectivity and sustained attention performance

Aversive smell associations shape social judgment

Contact Info

Product

Resources

About