Mind reading or delving into a person’s conscious experience is a concept only science fiction consists of. Though it sounds improbable, recent strides in neuroimaging have made it a reality, at least partially.

Visual stimuli are captured through the eyes and relayed into the brain for perception. Decoding this process involves a systemic mapping between the visual stimuli and brain activity. Scientists have previously used brain activity patterns from functional magnetic resonance imaging (fMRI) technique for reconstruction of images as perceived by the cortex of the brain. The signals were reconstructed through a decoder ‘trained’ by feeding patterns of known images. But, the resolution and accuracy of the reconstructed images was low.

A new method uses sophisticated decoding techniques to improve the quality and scope of the reconstructed images. For the experiment, artificial images, defined entirely by differences in local contrast were used as visual stimulus and presented to experimental subjects while fMRI signals were simultaneously recorded from V1, V2 and V3 areas of the cortex known as the early visual areas. Images were decoded using complicated overlapping regions and multiple scales. Relevant neural signals were automatically identified through the information represented in correlated activity patterns rather than mere use of known functional anatomy. By using this new method for brain decoding, researchers were able to reconstruct a large variety of images using only several hundred random images to train the reconstruction model. Also, almost 100% correct identification of images was achieved which is an impressive improvement from earlier methods.

As of now, none of the published research work claims reconstruction of natural images seen outside the laboratory in our daily life. This is difficult as we have only a rudimentary knowledge of how the higher visual areas of the cortex function.

This generalized procedure can be extended to deal with more complex perceptions consisting of numerous possible states by using multiple decoders. If a perceptual state can be expressed by a combination of elemental features, a modular decoder can be trained for each feature with a small number of data and their combination can accurately predict previously undetermined and novel perceptions.

Brain decoding has many potential applications and can also be used to decode olfactory, motor and sensory stimuli. Thus, a prosthetic device can be controlled by using signals from the motor cortex. It can be further extended to other mental states such as lie detection and even visualization of dreams. Though still in its infancy, this research direction raises uncomfortable questions about ethical issues concerning the privacy of human thought.

References

Kay K, & Gallant J (2009). I can see what you see Nature Neuroscience, 12 (3) DOI: 10.1038/nn0309-245

MIYAWAKI, Y., UCHIDA, H., YAMASHITA, O., SATO, M., MORITO, Y., TANABE, H., SADATO, N., & KAMITANI, Y. (2008). Visual Image Reconstruction from Human Brain Activity using a Combination of Multiscale Local Image Decoders Neuron, 60 (5), 915-929 DOI: 10.1016/j.neuron.2008.11.004