• Wayfarer
    13.8k
    Does a dog see a computer as a single entity?Joshs

    :up: Point well made.
  • Banno
    14.5k
    Yes; I'm well aware of such studies. Hence my note above, in which I made it clear that I think pattern recognition has a part to play.

    But it is not the whole story.

    Notice that the study I cited used both canonical and non-canonical spacial disposition.

    Thank you for citing something with a bit of empirical content... :wink:
  • Isaac
    5.5k


    The debate is still going on I'm afraid. I've not kept up with this thread, but as I understand it, the debate is about automaticity of number recognition, yes? The conflict between various stroop-like experiments manipulating visuo-spatial indicators of magnitude with actual numerical representations (a big number '7' vs a tiny number '9' - that sort of thing). Cohen Kadosh pretty much put that to bed in 2008, so papers from before then would have to be viewed in the light of a then open debate which is now considered less so. As any follower of developments in neuroscience will have come to expect, the matter turns out to be much more complex. Elements of visuo-spatial signals (size, pattern) are taken into account alongside priors expectations from things like ordinal and magnitude judgements in context. In fact Cohen Kadosh found that the specific instructions by the experimenters lead to different patterns of activation. Ultimately there's a significant degree of non-abstract numerical representation - numbers are represented in the brain not as the concept 'three' but as a combination of patterns, language, magnitude, numerosity, and even more synaesthetic relations. We know from something as simple as digit recognition that the inferotemporal layer can make 'best guesses' with ambiguous signals from the v4 regions (higher order perceptual features like overall shape, texture etc). It's likely that the final behaviour (assigning a number word, performing a calculation, weighing magnitudes...) is both dependant on, and influences (by backward acting suppression) the balance of collected 'evidence' in the v4 region.

    Basically, a range of evidence is gathered and which evidence takes priority is dependant on the task at hand. (don't know why I didn't just say that at the beginning, still...it's written now.)

    What I would add though, is that the often touted studies on infant number recognition are being misused to support mathematical realism. Infant studies done thus far just about all support the prevailing magnitude-estimation hypothesis, they are not about infants recognising 'the eternal number three' or any such. Magnitude estimation and numerosity are two different processing streams and shouldn't be confused. One can estimate the relative magnitude of two pages full of dots without having to, or even being able to, count them. Fine-scale magnitude recognition is both granular (ie based on individuating objects) and scalar. It's this granular magnitude recognition that's often misquoted as support for innate number recognition, but experiments such as this one https://pubmed.ncbi.nlm.nih.gov/11814309/ (and other subsequently) demonstrate that it is granular magnitude-recognition that underlies these infant abilities.
  • Banno
    14.5k
    Cheers. Just thought you might be interested, it being your area.

    In fact Cohen Kadosh found that the specific instructions by the experimenters lead to different patterns of activation.Isaac
    Complexity increasing with yet another loop.

    ...numbers are represented in the brain not as the concept 'three' but as a combination of patterns, language, magnitude, numerosity, and even more synaesthetic relations.Isaac

    Which in my ignorant head harmonises with the Wittgensteinian rejection of concepts as pieces of furniture in minds. If we are to look to use instead of meaning, we would not expect to find a "spot" in the brain for each number, but instead to see something reflecting the range of stuff we can do with numbers.

    Magnitude estimation and numerosity are two different processing streams and shouldn't be confused.Isaac
    Interesting then that Subitising seems, according to the study I mentioned, to use the same networks as counting - is that the same as numerosity?
  • Olivier5
    3.3k
    For those interested in how the abilities of counting or estimating magnitude pan out in other species than our own, Octolab has a series of fun experiments on octopuses, including on such topics as object permanence and numeration.

    https://octolab.tv/can-an-octopus-count-viewer-request/
1910111213Next
bold
italic
underline
strike
code
quote
ulist
image
url
mention
reveal
youtube
tweet
Add a Comment