Submitted via IRC for SoyCow4463
This brain region may be why some robots send chills down your spine
A new analysis of brain scans may explain why hyperrealistic androids and animated characters can be creepy.
By measuring people's neural activity as they viewed pictures of humans and robots, researchers identified a region of the brain that seems to underlie the "uncanny valley" effect — the unsettling sensation sometimes caused by robots or animations that look almost, but not quite, human (SN Online: 11/22/13). Better understanding the neural circuitry that causes this feeling may help designers create less unnerving androids.
In research described online July 1 in the Journal of Neuroscience, neuroscientist Fabian Grabenhorst and colleagues took functional MRI scans of 21 volunteers during two activities. In each activity, participants viewed pictures of humans, humanoid robots of varying realism and — to simulate the appearance of hyperrealistic robots — "artificial humans," pictures of people whose features were slightly distorted through plastic surgery and photo editing.
[...] Brain scans revealed that activity in the ventromedial prefrontal cortex, or VMPFC — a region involved in making value judgments — mirrored participants' uncanny valley reactions. VMPFC activity was typically higher in response to more humanlike pictures, but dipped in response to artificial humans. That drop was most pronounced in people with the strongest dislike for artificial humans. Those findings suggest that this region of the brain underpins the uncanny valley sensation, the researchers say.
[...] If the VMPFC is responsible for generating the uncanny valley heebie-jeebies, that may be good news for android designers and animators. Social experiences can change how VMPFC reacts to certain situations, says Grabenhorst, of the University of Cambridge. So positive interactions with an initially creepy robot or avatar may make it less bothersome.
(Score: 2) by ikanreed on Tuesday July 09 2019, @04:41PM (2 children)
This exact class of paper has a huge reproducibility problem [nih.gov]
Some regions of the brain show low contrast-to-noise levels and can be falsely indicated easily. This paper's methodology section doesn't clarify the question of the ventromedial prefrontal cortex's overall testability. And their data look pretty noisy. You can't clearly partition samples in figure D into positive and negative cases. And their arbitrary selection of time window for sampling opens the the risk of accidental(or even purposeful) cherry picking and p-hacking.
Plausible result with underlying mechanism, but the data doesn't make a profound and irrefutable case for the hypothesis, and there's valid methodology questions left open.
(Score: 0) by Anonymous Coward on Tuesday July 09 2019, @06:07PM (1 child)
Yep, I was about to post something similar. Here's a useful-looking intro to the fMRI process,
https://www.nature.com/scitable/blog/brain-metrics/what_does_fmri_measure [nature.com]
It may be the best tool available at the moment, but it is still a monster pile driver, attempting to gently push in a tiny tack.
Hints: It does not measure actual nerves being activated. Further the resolution is a joke, "In a typical scan, each voxel might cover 3mm^3 of tissue, a volume that would encompass ~ 630,000 neurons in cortex.
(Score: 2) by c0lo on Wednesday July 10 2019, @02:44AM
See also [prefrontal.org]: Bennett et al. "Neural Correlates of Interspecies Perspective Taking in the Post-Mortem Atlantic Salmon: An Argument For Proper Multiple Comparisons Correction" Journal of Serendipitous and Unexpected Results, 2010.
https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford