Icelandic Vision Lab
LightNews LightNews
banner
icevislab.bsky.social
Icelandic Vision Lab
@icevislab.bsky.social
1.1K followers 740 following 85 posts
Icelandic Vision Lab focuses on all things visual with a major emphasis on higher-level or "cognitive" aspects of vision: visionlab.is This is a joint account run by IVL PIs (Ásgeirsson, Hansmann-Roth, Kristjánsson, Ólafsdóttir, Sigurdardottir).
Posts Replies Media Videos
icevislab.bsky.social
Please distribute: PhD Position in Visual Perception -- joint opportunity in Iceland and Switzerland #visionscience #cogsci #neuroskyence #psychscisky #phd #academia
May 27, 2025 at 2:41 PM
icevislab.bsky.social
How can they ALL be different?! Worst visual search task ever ☹️😖 #visionscience
Black socks
May 10, 2025 at 8:04 AM
icevislab.bsky.social
Some people just LOVE Gabor patches. Model: @antonlukashevich.bsky.social Maker: Dr. Tyurina #visionscience
Gabor patch earring
April 4, 2025 at 5:56 PM
icevislab.bsky.social
😵😵😵🌈🌈🌈
(I hope colorblind people find this funny and not offensive but then again they wouldn't know. This is Heida, blame me sorry.) #visionscience #psychscisky #neuroskyence #sciencememes
Shocked Pikachu Ishihara Plate
March 13, 2025 at 1:13 PM
icevislab.bsky.social
ENDILEGA DEILA

Viltu taka þátt í rannsókn? Þráður 🧵

Upplifir þú liti þegar þú horfir á bókstafi?

Ef þú ert eins og manneskjan til hægri þá upplifir þú fyrirbæri sem nefnist samskynjun.

Til að geta rannsakað fyrirbærið þurfum við gögn frá fólki MEÐ og ÁN samskynjunar. 1/5
February 26, 2025 at 1:57 PM
icevislab.bsky.social
Getting review requests.

(I still say yes sometimes, sucker that I am.)

#academia #science
"They pay me in woims"
February 18, 2025 at 8:38 AM
icevislab.bsky.social
We suggest that the visual properties approximated by two primary dimensions of object space are automatically extracted by the human visual system during an initial bottom-up sweep. Our results support the contribution of object space to human visual cognition. 13/13 - thread sharing appreciated -
Panel A: Partial correlation plot between face discrimination ability and average discrimination ability for stubby animate-looking birds and hands; axes show residuals after partialling out average discrimination ability for spiky animate-looking birds and hands. Panel B: Partial correlation plot between face discrimination ability and average discrimination ability for spiky animate-looking birds and hands; axes show residuals after partialling out average discrimination ability for stubby animate-looking birds and hands (targets per second).
February 13, 2025 at 11:27 AM
icevislab.bsky.social
Specific abilities for faces were additionally partially accounted for by their visual qualities as measured by location in object space. Face abilities were better predicted by people's abilities for discriminating stubby animate-looking objects than their abilities for other objects. 12/13
Multidimensional scaling was performed on the correlation matrix for performance (targets per second) on individual trials. This was done separately for (a) trials with novel objects and object-matched faces and (b) trials with familiar objects and object-matched faces. Each dot depicts a single trial, and the distance between two points is approximately equal to their dissimilarity. Trials are clustered based on our preregistered analysis pipeline: The faces cluster corresponds to trials contributing to the face discrimination dependent variable in Table 2, the stubby-animate objects cluster corresponds to trials contributing to factors added at Step 4 in Table 2, and the other objects cluster corresponds to trials contributing to factors added at Step 3 in Table 2. Dots are connected to their cluster’s barycenter via lines. Confidence ellipses for barycenters of the three clusters are shown, both for the 95% confidence level (light ellipses) and the 50% confidence level (dark ellipses).
February 13, 2025 at 11:27 AM
icevislab.bsky.social
Face discrimination ability was surprisingly well predicted by visual discrimination ability for miscellaneous other objects, in accordance with
individual differences in a general object perception factor o. All object perception abilities cluster. 11/13
O-factor in object perception
February 13, 2025 at 11:27 AM
icevislab.bsky.social
These stimuli were then used in a visual foraging task where a large sample of participants (N = 511) with greatly varying face processing abilities had to search for target objects among distractor objects. 10/13
Stimuli belonged to one of four quadrants of object space. They were used in a visual foraging task where people searched for targets among distractors.
February 13, 2025 at 11:27 AM
icevislab.bsky.social
We took objects+faces and projected them onto object space. Faces clustered in the stubby animate-looking quadrant of object space. We matched their object space coordinates with those of various other stubby animate-looking objects. We also included objects from the other three quadrants. 9/13
Object space coordinates of stimuli (11 exemplars of each type). Quadrants of object space are represented by different colors (green: spiky inanimate-looking; pink: stubby inanimate-looking; purple: spiky animate-looking; orange: stubby animate-looking; gray: all objects). Each dot represents object space coordinates of a particular exemplar object image. Exerc. eq. = exercise equipment.
February 13, 2025 at 11:27 AM
icevislab.bsky.social
We took images of various things and ran them through a neural network. We extracted layer activations of a deep layer, performed dimensionality reduction, and extracted two dimension that explained the greatest amount of variance in the appearance of these objects. This was our object space. 8/13
The image demonstrates our overall approach to defining a visual object space. The depicted objects, network structure, activation patterns, and object space are purely illustrative. A convolutional neural network (VGG-16bn; Simonyan & Zisserman, 2014; see also Muttenthaler & Hebart, 2021) pretrained on ecologically relevant image categories (ecoset; Mehrer et al., 2017, 2021) was used to extract deep layer activations for a reference imageset (large and small animate and inanimate objects; Konkle & Caramazza, 2013). Principal components analysis (PCA) on these activations was used to define an object space that should capture the main two dimensions on which the reference objects vary in appearance. Network activation patterns of other objects, not used to construct object space, were then projected to this object space to extract their coordinates. A subset of these objects was used in the experiment.
February 13, 2025 at 11:27 AM
icevislab.bsky.social
However, this selectivity may reflect visual properties that these objects happen to have; faces are e.g. stubby animate-looking objects (fig. Bao et al. 2020). The ventral stream may be organized around a visual object space that captures the main dimensions on which objects visually differ. 4/13
Figure from Bao, She, McGill, & Tsao, 2020. Example images from the quadrants of object space.
February 13, 2025 at 11:27 AM
icevislab.bsky.social
Specific regions in the ventral visual pathway respond selectively to certain categories, including faces (fig. Sigurdardottir et al. 2021). 3/13
Topographical organization of the high-level ventral visual stream.
February 13, 2025 at 11:27 AM
icevislab.bsky.social
Unlike differences in other cognitive traits (e.g., Big Five personality traits), we have limited knowledge on how people differ in their object processing capacity and whether such abilities are fractionated or unitary. Faces however arguably have the greatest claim to domain specificity. 2/13
Left: Big Five personality factors. Right: Big X (unknown) of object perception.
February 13, 2025 at 11:27 AM
icevislab.bsky.social
Paper🚨 "Objects, Faces, and Spaces: Organizational Principles of Visual Object Perception as Evidenced by Individual Differences in Behavior" by @heidasigurdar.bsky.social & @ingamariao.bsky.social JEP:G editor's choice ->free to read psycnet.apa.org/fulltext/202... #visionscience #psychscisky 🧵1/13
Example stimuli of 15 object types are shown. Quadrants of object space are represented by different colors (orange: stubby animate-looking; purple: spiky animate-looking; pink: stubby inanimate-looking; green: spiky inanimate-looking; gray: all objects). Exerc. eq. = exercise equipment.
February 13, 2025 at 11:27 AM
icevislab.bsky.social
Two grants to IVL 🥳 "Serial effects in visual perception" to PI @shansmann-roth.bsky.social + "Haptic harmony" to postdoc Ivan Makarov in collab with Unnthorsson, Kristjánsson, Jónasson & Uenala. Sabrina is looking for candidates for a three-year Ph.D. and a two-year postdoc position: sabrina@hi.is
Northern lights over Iceland
February 11, 2025 at 6:11 PM
icevislab.bsky.social
Are we ready for this again? No we are not. #UnventEmails
every day i get emails
January 2, 2025 at 10:56 PM
icevislab.bsky.social
Results showed a robust link between VWM precision for simple stimuli and memory for configural information in houses, beyond faces/pseudowords. Expert categories (faces, words) reduce VWM demands, while non-expert categories (houses) impose greater demands, especially for configural processing. 4/6
The experimental design. On each trial, a sample image (face, house, or pseudoword) appeared in the middle of the screen, followed by match and foil images concurrently presented to the left and right of the center. The match was identical to the sample image; however, the foil was either featurally or configurally different. Match and foil pictures remained onscreen until the participant identified which image was identical to the sample image by pressing a button.
December 11, 2024 at 11:32 AM
icevislab.bsky.social
Additionally, they performed a delayed match-to-sample task involving faces, houses, and pseudowords, where individual features or feature configurations were manipulated. 3/6
Examples of featural or configural changes of faces, houses, and pseudowords used in the visual recognition task.
December 11, 2024 at 11:32 AM
icevislab.bsky.social
We explored the impact of visual working memory (VWM) constraints on the processing of complex objects, with a VWM task where participants (N = 75) adjusted the orientation of a bar to match a previously viewed one. 2/6
Visual working memory task. Four randomly oriented bars were displayed for 1000 ms, followed by a 1000 ms mask. Subsequently, one of the items in the array was probed by a line serving as a location post-cue and the participants were asked to rotate it to indicate the remembered orientation. Arrows illustrate orientation adjustment and were not actually shown around the adjustment target.
December 11, 2024 at 11:32 AM
icevislab.bsky.social
Paper alert! The impact of visual working memory constraints on object recognition www.tandfonline.com/eprint/KDXEF... #visionscience #cogsci #cognitivescience #cognitivepsychology #cogscisky Authors: @bahareh-jozranjbar.bsky.social, Kristjánsson, Gerlach, @heidasigurdar.bsky.social 🧵Thread 1/6
Network plot of recognition accuracy in the different conditions of the visual object recognition task and visual working memory (VWM) precision for simple stimuli. The network plot shows the pattern in Pearson’s correlations between the six object memory conditions and VWM precision. The color of the connections indicates the strength and direction of the correlations, with strong positive correlations shown in red shades and weak positive correlations in blue shade; there were no negative correlations. Thicker arcs represent stronger correlations. The graph's structure, obtained through multidimensional clustering, simplifies the high-dimensional correlation data for easier interpretation. Configural house recognition shows the strongest positive correlation with VWM precision, indicated by the dark red arc.
December 11, 2024 at 11:32 AM