Chris Racey
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Chris Racey has a total of 39 co-authors. Here are the top co-authors by number of shared publications:
- Charlotte L. Rae
University of Sussex, Brighton, UK
7 shared publications - Nicholas Edward Souter
University of Sussex, Brighton, UK
6 shared publications - Loïc Lannelongue
University of Cambridge
5 shared publications - Nikhil Bhagwat
McGill University
5 shared publications - Niall W. Duncan
Graduate Institute of Mind, Brain and Consciousness, Taipei Medical University, Taipei, Taiwan
4 shared publications Gabby Samuel
3 shared publications
Latest Publications
- Open Access
- 7 April 2026
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 123 (15)
The colors of images preferred by individual voxels can be used to delineate functionally distinct visually responsive brain areas
We exploited co-occurrences between color and other properties of natural scenes to identify and visualize functionally distinct brain regions. For each voxel in the Natural Scenes Dataset (NSD), we computed a scaled response-weighted average of the stimulus images. The colors of “voxel-preferred images” (VPIs) reflect stimulus properties that covary with color in natural scenes: Color serves as a tag for functional distinctions in voxel responses. Mapping VPIs onto cortical surfaces revealed reliable and structured color patterns that segment voxel clusters. Boundaries between clusters of similarly colored VPIs tend to coincide with boundaries defined using other methods, and heterogeneity within regions suggests functional subdivisions. VPIs provide a simple data-driven method for analyzing fMRI responses to natural scenes and visualizing cortical organization.
- 7 April 2026
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 123 (15)
Open AccessThe colors of images preferred by individual voxels can be used to delineate functionally distinct visually responsive brain areas
We exploited co-occurrences between color and other properties of natural scenes to identify and visualize functionally distinct brain regions. For each voxel in the Natural Scenes Dataset (NSD), we computed a scaled response-weighted average of the stimulus images. The colors of “voxel-preferred images” (VPIs) reflect stimulus properties that covary with color in natural scenes: Color serves as a tag for functional distinctions in voxel responses. Mapping VPIs onto cortical surfaces revealed reliable and structured color patterns that segment voxel clusters. Boundaries between clusters of similarly colored VPIs tend to coincide with boundaries defined using other methods, and heterogeneity within regions suggests functional subdivisions. VPIs provide a simple data-driven method for analyzing fMRI responses to natural scenes and visualizing cortical organization.
- Open Access
- 10 February 2026
- Published by Center for Open Science in PsyArXiv
fMRI biomarkers of length of working week: A UK Biobank study
Working longer hours is associated with negative outcomes in health and wellbeing. While effects of occupational stress on the brain have been observed, there have been no investigations into the effect of amount of time spent at work on brain function to date. We examined the effect of length of working week (in hours) on task-based and resting-state functional magnetic resonance imaging (fMRI) in 8,528 and 7,589 participants, respectively, from the UK Biobank. Analyses were conducted to treat working hours as both a continuous and a categorical (lower versus higher hours groups) variable. Covariates relevant to length of working week were selected using random forest machine learning and were systematically adjusted for in a sequential manner. No association with length of working week was found for task-based fMRI, as reflected by activation in the amygdala, fusiform face area, and insula during an emotional face matching task. For resting-state fMRI, greater continuous length of working week was found to be positively associated with within-network connectivity in the somatomotor network, but did not predict connectivity within or between any other network. The observed association may suggest compensatory or adaptive somatomotor connectivity changes in the face of increased job demands. All other within- and between-network connectivity metrics showed no significant relationship with working hours, following optimal model selection and correction for multiple statistical tests. When looking at uncorrected p-values, significant statistical associations with resting-state fMRI tended to become non-significant once demographic factors were adjusted for, suggesting that associations between length of working week and brain function are frequently accounted for by these covariates, with the sole exception of somatomotor connectivity. Overall, this suggests that length of working week, as just one employment factor, does not show strong associations with fMRI metrics, when related variables are adjusted for. Interventional studies, and analysis of additional factors such as burnout and shift work may shed further light on neural markers of occupational wellbeing.
- 10 February 2026
- Published by Center for Open Science in PsyArXiv
Open AccessfMRI biomarkers of length of working week: A UK Biobank study
Working longer hours is associated with negative outcomes in health and wellbeing. While effects of occupational stress on the brain have been observed, there have been no investigations into the effect of amount of time spent at work on brain function to date. We examined the effect of length of working week (in hours) on task-based and resting-state functional magnetic resonance imaging (fMRI) in 8,528 and 7,589 participants, respectively, from the UK Biobank. Analyses were conducted to treat working hours as both a continuous and a categorical (lower versus higher hours groups) variable. Covariates relevant to length of working week were selected using random forest machine learning and were systematically adjusted for in a sequential manner. No association with length of working week was found for task-based fMRI, as reflected by activation in the amygdala, fusiform face area, and insula during an emotional face matching task. For resting-state fMRI, greater continuous length of working week was found to be positively associated with within-network connectivity in the somatomotor network, but did not predict connectivity within or between any other network. The observed association may suggest compensatory or adaptive somatomotor connectivity changes in the face of increased job demands. All other within- and between-network connectivity metrics showed no significant relationship with working hours, following optimal model selection and correction for multiple statistical tests. When looking at uncorrected p-values, significant statistical associations with resting-state fMRI tended to become non-significant once demographic factors were adjusted for, suggesting that associations between length of working week and brain function are frequently accounted for by these covariates, with the sole exception of somatomotor connectivity. Overall, this suggests that length of working week, as just one employment factor, does not show strong associations with fMRI metrics, when related variables are adjusted for. Interventional studies, and analysis of additional factors such as burnout and shift work may shed further light on neural markers of occupational wellbeing.
- Open Access
- 16 June 2025
- Published by MIT Press in Imaging Neuroscience
- Vol. 3
Comparing the carbon footprint of fMRI data processing and analysis approaches
We compared the carbon emissions of preprocessing and statistical analysis of fMRI data in software packages FSL, SPM, and fMRIPrep using an existing open access dataset. Carbon emissions for fMRIPrep were 30x larger than those of FSL, and 23x those of SPM. We also compared the scientific performance of each package, reflected by sensitivity to statistical activation. Overall, fMRIPrep demonstrated slightly superior statistical sensitivity to both FSL and SPM, with FSL also outperforming SPM. However, this pattern varied by brain region. Researchers analysing fMRI data can use these findings to inform their choice of software package, considering the carbon footprint of data processing alongside usability and quality of derived output. Researchers should be conscious of how and when tools that elicit heavy compute are used, minimising energy usage and subsequent file size when possible. Researchers developing and using such tools should consider the extent to which computationally expensive steps are necessary to produce high-quality results.
- 16 June 2025
- Published by MIT Press in Imaging Neuroscience
- Vol. 3
Open AccessComparing the carbon footprint of fMRI data processing and analysis approaches
We compared the carbon emissions of preprocessing and statistical analysis of fMRI data in software packages FSL, SPM, and fMRIPrep using an existing open access dataset. Carbon emissions for fMRIPrep were 30x larger than those of FSL, and 23x those of SPM. We also compared the scientific performance of each package, reflected by sensitivity to statistical activation. Overall, fMRIPrep demonstrated slightly superior statistical sensitivity to both FSL and SPM, with FSL also outperforming SPM. However, this pattern varied by brain region. Researchers analysing fMRI data can use these findings to inform their choice of software package, considering the carbon footprint of data processing alongside usability and quality of derived output. Researchers should be conscious of how and when tools that elicit heavy compute are used, minimising energy usage and subsequent file size when possible. Researchers developing and using such tools should consider the extent to which computationally expensive steps are necessary to produce high-quality results.
- Open Access
- 9 April 2025
- Published by Center for Open Science in OSF Preprints
Comparing the carbon footprint of fMRI data processing and analysis approaches
We compared the carbon emissions of preprocessing and statistical analysis of fMRI data in software packages FSL, SPM, and fMRIPrep. Carbon emissions for fMRIPrep were 30x larger than those of FSL, and 23x those of SPM. We also compared the scientific performance of each package, reflected by sensitivity to statistical activation. Overall, fMRIPrep demonstrated slightly superior statistical sensitivity to both FSL and SPM, with FSL also outperforming SPM. However, this pattern varied by brain region. Researchers analysing fMRI data can use this information to inform their choice of software package, considering the carbon footprint of data processing alongside usability and quality of derived output. Researchers should be conscious of how and when tools that elicit heavy compute are used, minimising energy usage and subsequent file size when possible. Researchers developing and using such tools should consider the extent to which computationally expensive steps are necessary to produce high quality results.
- 9 April 2025
- Published by Center for Open Science in OSF Preprints
Open AccessComparing the carbon footprint of fMRI data processing and analysis approaches
We compared the carbon emissions of preprocessing and statistical analysis of fMRI data in software packages FSL, SPM, and fMRIPrep. Carbon emissions for fMRIPrep were 30x larger than those of FSL, and 23x those of SPM. We also compared the scientific performance of each package, reflected by sensitivity to statistical activation. Overall, fMRIPrep demonstrated slightly superior statistical sensitivity to both FSL and SPM, with FSL also outperforming SPM. However, this pattern varied by brain region. Researchers analysing fMRI data can use this information to inform their choice of software package, considering the carbon footprint of data processing alongside usability and quality of derived output. Researchers should be conscious of how and when tools that elicit heavy compute are used, minimising energy usage and subsequent file size when possible. Researchers developing and using such tools should consider the extent to which computationally expensive steps are necessary to produce high quality results.
- 6 March 2025
- Published by Optica Publishing Group in Journal of the Optical Society of America A
- Vol. 42 (5) , B335-B345
Comparison of intermodulation and oddball methods for measuring human cortical color tuning functions using steady-state visually evoked potentials
Characterization of postreceptoral chromatic mechanisms is an active challenge in color vision research. Adding to evidence from human psychophysics and primate neurophysiology, human neuroimaging using electroencephalography (EEG) in combination with steady-state visually evoked potentials (SSVEPs) has recently been used to investigate the tuning of cortical color mechanisms. We compare the performance of two SSVEP-based methods for measuring the tuning of cortical color mechanisms: an intermodulation component (IM)-based method and an oddball method. Our results support SSVEPs as an effective tool for investigating the tuning of cortical color mechanisms. The intermodulation method is efficient and provides high signal-to-noise ratios, but the oddball method can also yield clear tuning functions if multiple SSVEP harmonics are analyzed in combination.
- 6 March 2025
- Published by Optica Publishing Group in Journal of the Optical Society of America A
- Vol. 42 (5) , B335-B345
Comparison of intermodulation and oddball methods for measuring human cortical color tuning functions using steady-state visually evoked potentials
Characterization of postreceptoral chromatic mechanisms is an active challenge in color vision research. Adding to evidence from human psychophysics and primate neurophysiology, human neuroimaging using electroencephalography (EEG) in combination with steady-state visually evoked potentials (SSVEPs) has recently been used to investigate the tuning of cortical color mechanisms. We compare the performance of two SSVEP-based methods for measuring the tuning of cortical color mechanisms: an intermodulation component (IM)-based method and an oddball method. Our results support SSVEPs as an effective tool for investigating the tuning of cortical color mechanisms. The intermodulation method is efficient and provides high signal-to-noise ratios, but the oddball method can also yield clear tuning functions if multiple SSVEP harmonics are analyzed in combination.
- Open Access
- 15 August 2024
- Published by Wiley in Human Brain Mapping
- Vol. 45 (12) , e70003
Measuring and reducing the carbon footprint of fMRI preprocessing in fMRIPrep
Computationally expensive data processing in neuroimaging research places demands on energy consumption—and the resulting carbon emissions contribute to the climate crisis. We measured the carbon footprint of the functional magnetic resonance imaging (fMRI) preprocessing tool fMRIPrep, testing the effect of varying parameters on estimated carbon emissions and preprocessing performance. Performance was quantified using (a) statistical individual‐level task activation in regions of interest and (b) mean smoothness of preprocessed data. Eight variants of fMRIPrep were run with 257 participants who had completed an fMRI stop signal task (the same data also used in the original validation of fMRIPrep). Some variants led to substantial reductions in carbon emissions without sacrificing data quality: for instance, disabling FreeSurfer surface reconstruction reduced carbon emissions by 48%. We provide six recommendations for minimising emissions without compromising performance. By varying parameters and computational resources, neuroimagers can substantially reduce the carbon footprint of their preprocessing. This is one aspect of our research carbon footprint over which neuroimagers have control and agency to act upon.
- 15 August 2024
- Published by Wiley in Human Brain Mapping
- Vol. 45 (12) , e70003
Open AccessMeasuring and reducing the carbon footprint of fMRI preprocessing in fMRIPrep
Computationally expensive data processing in neuroimaging research places demands on energy consumption—and the resulting carbon emissions contribute to the climate crisis. We measured the carbon footprint of the functional magnetic resonance imaging (fMRI) preprocessing tool fMRIPrep, testing the effect of varying parameters on estimated carbon emissions and preprocessing performance. Performance was quantified using (a) statistical individual‐level task activation in regions of interest and (b) mean smoothness of preprocessed data. Eight variants of fMRIPrep were run with 257 participants who had completed an fMRI stop signal task (the same data also used in the original validation of fMRIPrep). Some variants led to substantial reductions in carbon emissions without sacrificing data quality: for instance, disabling FreeSurfer surface reconstruction reduced carbon emissions by 48%. We provide six recommendations for minimising emissions without compromising performance. By varying parameters and computational resources, neuroimagers can substantially reduce the carbon footprint of their preprocessing. This is one aspect of our research carbon footprint over which neuroimagers have control and agency to act upon.
- Open Access
- 19 January 2024
- Published by Center for Open Science in OSF Preprints
Measuring and reducing the carbon footprint of fMRI preprocessing in fMRIPrep
Computationally expensive data processing in neuroimaging research places demands on energy consumption – and the resulting carbon emissions contribute to the climate crisis. We measured the carbon footprint of the fMRI preprocessing tool fMRIPrep, testing the effect of varying parameters on estimated carbon emissions and preprocessing performance. Performance was quantified using (a) statistical individual-level task activation in regions of interest and (b) mean smoothness of preprocessed data. Eight variants of fMRIPrep were run with 257 participants who had completed an fMRI stop signal task (the same data also used in the original validation of fMRIPrep). Some variants led to substantial reductions in carbon emissions without sacrificing data quality: for instance, disabling FreeSurfer surface reconstruction reduced carbon emissions by 48%. We provide six recommendations for minimising emissions without compromising performance. By varying parameters and computational resources, neuroimagers can substantially reduce the carbon footprint of their preprocessing. This is one aspect of our research carbon footprint over which neuroimagers have control and agency to act upon.
- 19 January 2024
- Published by Center for Open Science in OSF Preprints
Open AccessMeasuring and reducing the carbon footprint of fMRI preprocessing in fMRIPrep
Computationally expensive data processing in neuroimaging research places demands on energy consumption – and the resulting carbon emissions contribute to the climate crisis. We measured the carbon footprint of the fMRI preprocessing tool fMRIPrep, testing the effect of varying parameters on estimated carbon emissions and preprocessing performance. Performance was quantified using (a) statistical individual-level task activation in regions of interest and (b) mean smoothness of preprocessed data. Eight variants of fMRIPrep were run with 257 participants who had completed an fMRI stop signal task (the same data also used in the original validation of fMRIPrep). Some variants led to substantial reductions in carbon emissions without sacrificing data quality: for instance, disabling FreeSurfer surface reconstruction reduced carbon emissions by 48%. We provide six recommendations for minimising emissions without compromising performance. By varying parameters and computational resources, neuroimagers can substantially reduce the carbon footprint of their preprocessing. This is one aspect of our research carbon footprint over which neuroimagers have control and agency to act upon.
- Open Access
Sensory representations in primary visual cortex are not sufficient for subjective imagery
The contemporary definition of mental imagery is characterized by two aspects: a sensory representation resembling, but not resulting from, perception, and an associated subjective experience. Neuroimaging demonstrated imagery-related sensory representations in primary visual cortex (V1) that show striking parallels to perception. However, it remains unclear whether these representations always reflect subjective experience, or they can be dissociated from it. We addressed this question by comparing sensory representations and subjective imagery among visualizers and aphantasics, the latter with an impaired ability to experience imagery. Importantly, to test for the presence of sensory representations independently of the ability to generate imagery on demand we examined both spontaneous and voluntary imagery forms. Using multivariate fMRI, we tested for decodable sensory representations in V1 and subjective visual imagery reports that occurred either spontaneously (during passive listening of evocative sounds) or in response to the instruction to voluntarily generate imagery of the sound content (always while blindfolded inside the scanner). Among aphantasics, V1 decoding of sound content was at chance during voluntary imagery, and lower than in visualizers, but it succeeded during passive listening, despite them reporting no imagery. In contrast, in visualizers, decoding accuracy in V1 was greater in voluntary than spontaneous imagery (while being positively associated with the reported vividness of both imagery types). Finally, for both conditions, decoding in precuneus was successful in visualizers but at chance for aphantasics. Together, our findings show that V1 representations can be dissociated from subjective imagery, while implicating a key role of precuneus in the latter.
Open AccessSensory representations in primary visual cortex are not sufficient for subjective imagery
The contemporary definition of mental imagery is characterized by two aspects: a sensory representation resembling, but not resulting from, perception, and an associated subjective experience. Neuroimaging demonstrated imagery-related sensory representations in primary visual cortex (V1) that show striking parallels to perception. However, it remains unclear whether these representations always reflect subjective experience, or they can be dissociated from it. We addressed this question by comparing sensory representations and subjective imagery among visualizers and aphantasics, the latter with an impaired ability to experience imagery. Importantly, to test for the presence of sensory representations independently of the ability to generate imagery on demand we examined both spontaneous and voluntary imagery forms. Using multivariate fMRI, we tested for decodable sensory representations in V1 and subjective visual imagery reports that occurred either spontaneously (during passive listening of evocative sounds) or in response to the instruction to voluntarily generate imagery of the sound content (always while blindfolded inside the scanner). Among aphantasics, V1 decoding of sound content was at chance during voluntary imagery, and lower than in visualizers, but it succeeded during passive listening, despite them reporting no imagery. In contrast, in visualizers, decoding accuracy in V1 was greater in voluntary than spontaneous imagery (while being positively associated with the reported vividness of both imagery types). Finally, for both conditions, decoding in precuneus was successful in visualizers but at chance for aphantasics. Together, our findings show that V1 representations can be dissociated from subjective imagery, while implicating a key role of precuneus in the latter.
- Open Access
- 8 August 2023
- Published by Center for Open Science in OSF Preprints
Ten recommendations for reducing the carbon footprint of research computing in human neuroimaging
Given that scientific practices contribute to the climate crisis, scientists should reflect on the planetary impact of their work. Research computing can have a substantial carbon footprint in cases where researchers employ computationally expensive processes with large amounts of data. Analysis of human neuroimaging data, such as Magnetic Resonance Imaging brain scans, is one such case. Here, we consider ten ways in which those who conduct human neuroimaging research can reduce the carbon footprint of their research computing, by making adjustments to the ways in which studies are planned, executed, and analysed; as well as where and how data is stored.
- 8 August 2023
- Published by Center for Open Science in OSF Preprints
Open AccessTen recommendations for reducing the carbon footprint of research computing in human neuroimaging
Given that scientific practices contribute to the climate crisis, scientists should reflect on the planetary impact of their work. Research computing can have a substantial carbon footprint in cases where researchers employ computationally expensive processes with large amounts of data. Analysis of human neuroimaging data, such as Magnetic Resonance Imaging brain scans, is one such case. Here, we consider ten ways in which those who conduct human neuroimaging research can reduce the carbon footprint of their research computing, by making adjustments to the ways in which studies are planned, executed, and analysed; as well as where and how data is stored.
- Open Access
- 23 July 2023
- Published by Center for Open Science in PsyArXiv
Can the Neural Representation of Physical Pain Predict Empathy for Pain in Others?
The question of whether physical pain and vicarious pain have some shared neural substrates is unresolved. Whilst most studies have shown overlapping regions of brain activity to these two states this need not reflect shared substrates at the voxel level or below. Krishnan et al. (2016) found that physical and vicarious pain are represented by dissociable multivariate brain patterns by creating biomarkers for physical pain (Neurologic Pain Signature, NPS) and vicarious pain (Vicarious Pain Signature, VPS) respectively. In the current research, the NPS and VPS were applied to three fMRI datasets (one new, two published) relating to vicarious pain which focused on between-subject differences in vicarious pain (Datasets 1 and 3, N = 40 and 65 participants) and within-subject manipulations of perspective taking (Dataset 2,N= 71 participants). Results show that (1) NPS can distinguish brain responses to images of pain versus no-pain and to a greater extent in vicarious pain responders who report experiencing pain when observing pain, (2) VPS can distinguish brain responses to images of pain versus no-pain only during self-perspective instructions. This study suggests that the NPS (created to detect physical pain) is, under some circumstances, sensitive to vicarious pain and that the VPS (created to detect vicarious pain) is not a generalizable biomarker of vicarious pain.
- 23 July 2023
- Published by Center for Open Science in PsyArXiv
Open AccessCan the Neural Representation of Physical Pain Predict Empathy for Pain in Others?
The question of whether physical pain and vicarious pain have some shared neural substrates is unresolved. Whilst most studies have shown overlapping regions of brain activity to these two states this need not reflect shared substrates at the voxel level or below. Krishnan et al. (2016) found that physical and vicarious pain are represented by dissociable multivariate brain patterns by creating biomarkers for physical pain (Neurologic Pain Signature, NPS) and vicarious pain (Vicarious Pain Signature, VPS) respectively. In the current research, the NPS and VPS were applied to three fMRI datasets (one new, two published) relating to vicarious pain which focused on between-subject differences in vicarious pain (Datasets 1 and 3, N = 40 and 65 participants) and within-subject manipulations of perspective taking (Dataset 2,N= 71 participants). Results show that (1) NPS can distinguish brain responses to images of pain versus no-pain and to a greater extent in vicarious pain responders who report experiencing pain when observing pain, (2) VPS can distinguish brain responses to images of pain versus no-pain only during self-perspective instructions. This study suggests that the NPS (created to detect physical pain) is, under some circumstances, sensitive to vicarious pain and that the VPS (created to detect vicarious pain) is not a generalizable biomarker of vicarious pain.