ben harvey

@_SidVerma Nah, totes funge

The pound is now a non-fungible token of Britain’s imperial past

1/3 I am recruiting graduate students interested in understanding the cognitive and neural mechanisms that support mathematical cognition and executive functions and how those relate to math development. Join me in Nashville at Vanderbilt! Due Dec. 1st . Reach out if interested!

@MartinMwiener Either of Ted Chiang's books, Exhalation and "Stories of Your Life and Others". Both incredible, page turners, with a slight sci-fi edge.

@mjhayashi @evi_hendrikx @Jacob_M_Paul @NathanvdStoep Thanks Michi, some nice relationships to your adaptation and mvpa studies!

@benmarkharvey @evi_hendrikx @Jacob_M_Paul @NathanvdStoep Nice work! Congrats!!

Our new paper at Nature Communications shows how visual timing-tuned neural responses emerge gradually through the visual hierarchy and abstract temporal responses from spatial responses. With @evi_hendrikx @Jacob_M_Paul Martijn van Ackooij @NathanvdStoep doi.org/10.1038/s41467…

In our new article w/Shir Hofstter @benmarkharvey @serge_dumoulin in Cell Rep, we ask how attention affects neural numerosity-selective response and find that attention is necessary to numerosity response, while without attention, response to preferred numerosity is suppressed.

With Martijn van Ackooij, @Jacob_M_Paul Wietske van der Zwaag and @NathanvdStoep

Our new article describes tuned responses to the duration and rate of auditory events in the human brain using 7T fMRI. While these responses are similar to those seen for visual event timing, the brain areas involved are very different: auditory cortices. doi.org/10.1016/j.neur…

@sami_r_yousif @Jacob_M_Paul This is an exciting topic for further research. Once we have an idea of the steps involved in numerosity estimation, we can look at those. It's better than continuing to discuss non-numerical image features without a computational model.

@sami_r_yousif @Jacob_M_Paul True. Numerosity perception shows a number of effects where true and perceived numerosity differ. We discuss some that might arise by the spatial frequency image representation, others that might arise in conversation to numerosity selective responses, and also high-level effects

Numerosity is estimated directly in early visual cortex via aggregate Fourier power - a satisfying solution to a super rewarding 7T fMRI postdoc project with @benmarkharvey, keen to discuss with everyone at #MCLS2022 @mathcogsociety

@sami_r_yousif @Jacob_M_Paul @mathcogsociety A phase scrambled image might seem like a good way to test this, but it's mostly useless when V2's response requires phase coherence.

@benmarkharvey @Jacob_M_Paul @mathcogsociety Got it. But if this is a *mechanism* of number perception, is it odd that they can be so easily disentangled? I'm having trouble understanding what else you're suggesting prevents me from encoding that phase-scrambled image as having the same number.

@sami_r_yousif @Jacob_M_Paul @mathcogsociety So V1 responds to Fourier power rather than numerosity, but later stages probably rely on phase coherence in estimating numerosity.

@sami_r_yousif @Jacob_M_Paul @mathcogsociety We discuss this in the paper. It is very simple to use phase scrambling to make an image with exactly the same Fourier spectrum, but with the phases at different frequencies out of alignment. While V1 responds well to phase scrambled images, V2 does not.

But instead of estimating numerosity directly from images, we estimate numerosity from the early visual representation of images. This is so straightforward in a spatial frequency based image representation that any animal with such spatial frequency processing could do the same.

Ever since I started thinking about numerosity, I have wondered how the visual system estimates numerosity so easily without large effects of item size and spacing. Our new paper at Nature Communication with @Jacob_M_Paul shows how this could be done. nature.com/articles/s4146…

Because it has been unclear how numerosity could be estimated from the visual image, many researchers have concluded that we don't really perceive numerosity, but some non-numerical feature that might be easier to estimate. Here we propose that we do truly estimate numerosity.

This explains why neural population response amplitudes in early visual cortex and neural networks follow numerosity so closely: they follow Fourier power more closely. But from this early visual representation, truly numerosity selective neural responses can be easily derived.

Animal visual systems use a spatial frequency based image representation. In the spatial frequency domain, numerosity becomes very clear: Fourier power at all orientations and spatial frequencies follows numerosity closely (but nonlinearly) regardless of item size and spacing.

@jordicat I don't think I perceive number, size or duration in my sense of taste or smell. So there is probably no neural representation of quantities in gustation or olfaction, though vestibular sense is less clear. I have enough to study in vision, hearing, motor and tactile processing!

@benmarkharvey Is the possibility worth exploring?

Our new paper is just out at Nature Communications. It uses 7T fMRI to show neural populations tuned for haptic numerosity, the number of items explored by touch. These are topographically mapped. They are distinct from visual numerosity maps, but overlap. nature.com/articles/s4146…