4 ms·

Did I miss the link to the original puplication, or did they actually fail to include it? http://www.nature.com/neuro/journal/vaop/ncurrent/full/nn.2795.html h

• jerf 16y ago
  Why would you need more than 5? This isn't a drug trial trying to pick up some subtle difference in results between taking a drug and not taking a drug. Is it really plausible that a theoretical person born blind and sighted later in fact can distinguish between the two but we just happened to get unlucky to find the five dunces in the world who can't? Based on what we already know about how the brain works (aka "the prior probabilities") the chances of such an outcome are pretty much insignificant, far lower than a naive sample-based argument that ignores what we already know would show.
  • perlgeek 16y ago
    > Why would you need more than 5? because humans are quite diverse > Is it really plausible that a theoretical person born blind and sighted later in fact can distinguish between the two but we just happened to get unlucky to find the five dunces in the world who can't? We don't know which percentage of blind people can distinguish objects by sight once they start to see. We don't know if it's all-or-nothing or not. If half of blinds can, there's still a 3% chance to get the published results from five subjects. If it's 30%, the chances grow to 17%.