The goal of my research is to understand the relationship between human perception and neural circuitry. Human visual cortex is organised into a hierarchy of discrete cortical maps, in which nearby neurons have receptive fields encoding information from nearby regions of visual space, forming a map of the visual field (i.e., a retinotopic map). I really care about retinotopic maps. I'm taken back when I find out somebody doesn't really care about them. I've always liked maps. I have framed old maps of cities on my apartment walls. I think it is because I like the idea of being able to quantify space. I like old clocks too for the same reason, but for time. Space and time are important, so we need to measure them! There is variability in the organisation of these retinotopic maps –in the distribution of cortical tissue representing different location of the visual field within a map, and in the overall surface area of the maps when comparing between individuals. There is up to 3-fold variability in the overall size of V1. This means that if you take a sample of V1s –say 30 of them– the largest V1 will have 3x the surface area when compared to the smallest V1. That is a huge amount of variation. And it's not related to the overall size of the brain which isn't very variable at all (everyone asks that). I am interseted in where this map variability comes from, and what it means for how we process and perceive the world around us.

My research combines fMRI, psychophysics, and computational modelling to study within- and between-subject variability in the organization of human retinotopic maps (focusing on primary visual cortex; V1, arguably the most important map!) with the goal of understanding how such variability impacts human visual perception.

Human visual performance changes with visual field location (see Himmelberg et al. 2023, Trends in Neurosciences for a lit review on this). It is well known that visual performance for most tasks (but not all!) is best at the center of the visual field and declines with eccentricity (i.e., distance toward the periphery). This eccentricity-dependent variability is consistent with V1 cortical magnification –the amount of cortical tissue (mm2) encoding 1° of visual space– which also declines with eccentricity. It's also consistent with the density of photoreceptors and retinal ganglion cells. Much less well-known is that visual performance varies markedly with polar angle –the position of a stimulus around a 360° of angle. Visual performance for a wide range of tasks is better along the horizontal than vertical, and lower than upper vertical meridian of the visual field. The differences are pretty substantial, we are way way better along the horizontal than vertical meridian. In fact, on some tasks people perform completely at chance on the upper vertical meridian but do perfectly fine on the horizontal. We are really bad on the upper vertical, but you would never know without actually testing it. My research has harnessed these polar angle asymmetries to better understand the relationship between cortical anatomy, neural encoding, and visual perception. I am going to detail just a few of the key studies below that link nicely, but there's a handful of related studies you could find on my google scholar if you happen to find any of this work interesting.

First, we used psychophysics to detail how contrast sensitivity –the currency of the visual system (my phd advisor said this to me once and I've used it ever since, I have no idea where he got it from. If I google it I get my own papers... so maybe it was him)– varies with polar angle location in the visual field, and between individual observers. I like measuring contrast sensitivity. I like that it is so fundamental to vision and I like that it is dictated by V1 neurons to a large extent. We confirmed polar angle asymmetries for contrast sensitivity. The asymmetries in contrast sensitivity persisted across manipulations of three stimulus parameters that dramatically alter contrast sensitivity; stimulus eccentricity, spatial frequency, and size (Himmelberg et al. 2020, Journal of Vision). This suggested that there must be asymmetries at multiple stages of the visual system. Critically, the extent of the asymmetries varied between individuals; some observers had far greater contrast sensitivity along the horizontal than the vertical meridian, whereas others this asymmetry was much weaker. In short, the extent of the asymmtery varied between individuals. This might mean that there is variability in their neural substrates (which we figured was probably going to be cortical magnification). We then sought to test the neural substrates of these two forms of variability.

Next, we showed that variability in contrast sensitivity –between locations and between observers– is linked to cortical magnification. In one study, we used fMRI to identify polar angle asymmetries in V1 cortical magnification; greater V1 tissue represents the horizontal than vertical, and lower than upper vertical meridian of the visual field, in parallel to the perceptual asymmetries (Himmelberg et al, 2021, NeuroImage). Noah found this first in his eLife paper though to be clear. We just reproduced the finding in a different dataset. Following this, we combined fMRI and psychophysical measures in the same observers to show that: 1) When individuals have a larger V1, they have greater overall contrast sensitivity; 2) when observers have more V1 tissue encoding a polar angle location in the visual field, they have greater contrast sensitivity at that location (relative to someone with less tissue encoding the same location) (Himmelberg et al., 2022, Nature Communications). We were chuffed. Virsu and Rovamo had a linking hypothesis all the way back in 1979 that said contrast sensitivity should be determined by the number of activated V1 neurons. And it looks like it's probably correct; more surface area = more neurons = higher contrast sensitivity. This project established a link between variability in the organisation of V1 and visual perception, both at the level of stimulus location and individual observer.

Following this, we harnessed the polar angle asymmetries to identify developmental changes in how V1 samples visual space between childhood and adulthood (Himmelberg et al, 2023, Nature Communications). Again using fMRI, we found that, children (ages 5-12) have a similar amount of V1 tissue representing the lower and upper vertical meridians of the visual field. This differs from adults, who have more V1 tissue representing the lower than upper vertical meridian. These brain measurements match visual performance in both groups; unlike adults, children’s visual performance is similar between the lower and upper vertical meridians (this paper was tested on like 200 kids or something crazy). So we realised that: 1) the asymmtery between the lower and upper vertical meridian has to emerge between childhood and adulthood; and 2) there are changes in V1 tissue/RF properties along the lower vertical meridian could drive the emergence of the asymmtery. The problem is though we don't know what the biological underpinning is. We aren't 'growing more neurons' but something has to be changing, don't know what though. I liked this paper because the most convincing results (to me) were just... looking all the childrens V1 maps, and all the adults maps. The difference between the groups was so clear. Overall though, this late stage change in the organisation of V1 was surprising, given that many properties of primary sensory areas develop and mature early in life. I should note that this is a project we did in collaboration with Stanford, where the data were collected. I don't even know where to start if I had to collect fMRI data from 5 year olds.

Together, these findings have been able to link visual perception to cortical magnification; within-subject variability in V1 cortical magnification drives some –but not all– variability in visual perception as a function of visual field location, and between-subject variability in V1 cortical magnification drives individual variability in visual perception. And then we were able to use the asymmetries to look at how V1 develops beyond childhood.

That's roughly what I've done during my post-doc. My current projects involve building a computational model of V1 that is constrained by multiple fMRI measurements of V1 properties with the goal of predicting visual performance throughout the visual field. If we know an individuals V1 cortical magnification/SF tuning/ orientation tuning/pRF size for some location in the visual field, can we predict that observers perceptual measurements at that location? It's complicated. I am also exploring co-variability in cortical magnification and the neural encoding of spatial frequency information; so linking cortical structure to neural encoding rather than cortical structure to visual perception. I'll spare you from my future plans.

A bit about me outside of research is that I am an avid casual amatuer powerlifter (my 1RM lifts are 280 bench/365 squat/415 conventional deadlift). I am also a 3x Hall of Famer at my local dive bar, a fan of Neon Genesis Evangelion –and more recently Demon Slayer– and spend a lot of my time lurching around the East Village in Manhattan. I also like to write and read. I play a lot of Fortnite too. Now I'm oversharing... I have been trying to think of ways to combine these interests into my research but I haven't been able to figure anything out so far :(. I also grew up in Sydney, Australia, then lived in the UK for roughly 8 years, and now I live in New York City.

Google Scholar
ORCID iD iconOrchidID