fMRI Mind ReadingCan fMRI read your mind?
|
|
Traditional fMRI experiments typically employ forward inference. For example, a subject might be shown pictures of a dog while fMRI determines which voxels are most strongly activated (e.g., voxels 15/19/47). fMRI mind reading inverts this relationship requiring a reverse inference to be made. In other words, if fMRI shows activation of voxels 15/19/47 can we conclude the subject is seeing a dog?
Several groups have tackled such problems using Multi-Voxel Pattern Analysis (MVPA) and artificial intelligence methods with machine learning. Jack Gallant's team from the University of California at Berkley have been perhaps the most successful allowing visual decoding complex scenes and even creation of a brain "dictionary".
|
|
|
Several commercial enterprises have jumped on the potential of fMRI for mind reading, including its use for "neuro-marketing" (e.g., determining brand preferences or response to commercials) and even lie detection (based on work by Harvard's Joshua Greene that the dorsolateral prefrontal cortex is more active in subjects contemplating lying).
As discussed in the videos below by psychologist Tor Wager from the University of Colorado at Boulder, reverse inference experiments are tricky to do well and are prone to misinterpretation. Thus I believe that current commercial applications of fMRI are premature and their claims should be viewed with great skepticism. If performed and properly analyzed, however, fMRI can read your mind to a limited (but surprising) extent, with many exciting developments expected for the future.
|
|
|
References
De Martino F, Valente G, Staeren N, et al. Combining multivariate voxel selection and
support vector machines for mapping and classification of fMRI spatial patterns. NeuroImage 2008; 43:44–58.
Greene JD, Paxton JM. Patterns of neural activity associated with honest and dishonest moral decisions. Proc Nat Acad Sci (USA) 2009; 106:12506-12511. (fMRI as a lie detector?)
Hassabis D, Chu C, Rees G, et al. Decoding neuronal ensembles in the human hippocampus. Curr Biol 2009; 19:546-554. (successfully used fMRI to predict the location of a subject moving between rooms in a virtual reality simulation)
Huth AG, de Heer WA, Griffiths TL, Theunissen FE, Gallant JL. Natural speech reveals the semantic maps that tile human cerebral cortex. Nature 2016; 532:453-458.
Huth AG, Nishimoto S, Vu AT, Gallant JL. A continuous semantic space describes the representation of thousands of object and action categories across the human brain. Neuron 2012; 76:1210-1224.
Kosslyn SM. If neuroimaging is the answer, what is the question? Phil Trans R Soc Lond B 1999; 354:1283-1294. (classic paper discussing limitations on the types of inferences and conclusions possible from fMRI and similar experiments)
Mahmoudi A, Takerkart S, Regragui F, et al. Multivoxel pattern analysis for fMRI data: a review. Comput Math Methods Med 2012: article ID 961257:1-14.
Pereira F, Mitchell T, Botvinick M. Machine learning classifiers and fMRI: a tutorial overview. NeuroImage 2009; 45:S199–209.
Poldrack RA. Can cognitive processes be inferred from neuroimaging data? Trends Cogn Sci 2006; 10:59–63. (describes forward and reverse inference concepts)
De Martino F, Valente G, Staeren N, et al. Combining multivariate voxel selection and
support vector machines for mapping and classification of fMRI spatial patterns. NeuroImage 2008; 43:44–58.
Greene JD, Paxton JM. Patterns of neural activity associated with honest and dishonest moral decisions. Proc Nat Acad Sci (USA) 2009; 106:12506-12511. (fMRI as a lie detector?)
Hassabis D, Chu C, Rees G, et al. Decoding neuronal ensembles in the human hippocampus. Curr Biol 2009; 19:546-554. (successfully used fMRI to predict the location of a subject moving between rooms in a virtual reality simulation)
Huth AG, de Heer WA, Griffiths TL, Theunissen FE, Gallant JL. Natural speech reveals the semantic maps that tile human cerebral cortex. Nature 2016; 532:453-458.
Huth AG, Nishimoto S, Vu AT, Gallant JL. A continuous semantic space describes the representation of thousands of object and action categories across the human brain. Neuron 2012; 76:1210-1224.
Kosslyn SM. If neuroimaging is the answer, what is the question? Phil Trans R Soc Lond B 1999; 354:1283-1294. (classic paper discussing limitations on the types of inferences and conclusions possible from fMRI and similar experiments)
Mahmoudi A, Takerkart S, Regragui F, et al. Multivoxel pattern analysis for fMRI data: a review. Comput Math Methods Med 2012: article ID 961257:1-14.
Pereira F, Mitchell T, Botvinick M. Machine learning classifiers and fMRI: a tutorial overview. NeuroImage 2009; 45:S199–209.
Poldrack RA. Can cognitive processes be inferred from neuroimaging data? Trends Cogn Sci 2006; 10:59–63. (describes forward and reverse inference concepts)
Related Questions
How do you statistically analyze fMRI data?
How is network theory used to analyze fMRI data?
How do you statistically analyze fMRI data?
How is network theory used to analyze fMRI data?