University of Trieste · Department of Life Sciences
A day of talks on perception and cognition, since 1993.
TSPC brings together researchers of visual perception, attention and cognitive processes for a day of oral communications, hosted in Trieste. Traditionally, the Symposium is open to all perspectives and approaches, has no registration fee, and runs on an informal, relaxed pace. This year, it is organized together with colleagues from the University of Padua.
Upcoming edition · 2026
As every year, the day opens with oral communications by researchers on perception, attention and cognition, continues with a poster session, and closes with the Kanizsa Lecture.
Oral communications — Room 1_C
11 talks (18 minutes each). Two equipment presentations (ULTRA, NIRx/Artinis) run as short breaks within the session.
Venue: Building F2, Room 1_C, first floor, San Giovanni campus, Via Edoardo Weiss 9, Trieste · map ↗
Lorenzo Tausani — SISSA, Trieste
Hierarchical processing of natural images along rat lateral extrastriate cortex: the case for a rodent ventral-like visual pathway
Unlike the primate ventral stream, the rodent visual hierarchy has traditionally been considered a shallow system capable of processing mainly low-level image features — a view reinforced by the failure of mouse studies to identify a clear functional specialization for cortical processing of visual objects. Yet this conclusion sits at odds with evidence of hierarchical shape processing in rat lateral extrastriate cortex, where deeper areas support recognition of visual objects despite changes in position, size and viewpoint, a landmark of primate object vision — though this evidence relied on simplified parametric shapes, leaving open whether the hierarchical organization persists with more naturalistic images. Neuropixels recordings from four rat visual cortical areas (V1, LM, LI, LL) in anesthetized rats repeatedly presented with 992 natural images spanning 62 ImageNet categories showed that encoding of category information in higher-order areas depends less on image luminance than in V1, consistent with a progression toward representations less dominated by low-level attributes; once luminance cues were accounted for, a clear, progressive increase in the ability of neuronal populations to discriminate image categories emerged from V1 to LL. Comparing image representations along the rat's lateral areas to successive layers of convolutional neural networks (CNNs) trained on ImageNet classification showed that activity in progressively higher-order cortical areas was best predicted by increasingly deeper CNN layers — a correspondence also found, applying the same pipeline, in public macaque datasets, but not in mouse datasets. These findings suggest that the functional structure of visual representations in rats may more closely resemble that of primates than that of mice, pointing to the rat as a complementary species for the study of object vision.
Chiara Di Domenico — SISSA, Trieste
A population signature of efficient coding for natural image statistics in rat visual cortex
Efficient coding theory posits that perceptual systems maximize encoded information while minimizing encoding costs by tuning neural representations to natural signal statistics. In vision, natural signals are characterized by local spatial correlations in light distribution, known as multipoint correlations, whose variability across natural scenes follows a characteristic ordering: 2-point correlations vary more than 4-point correlations, which in turn vary more than 3-point correlations — and perceptual sensitivity in both humans and rats follows the same ranking, consistent with an efficient adaptation of mammalian visual systems to natural texture statistics. Yet it remains unknown where a neural signature of such efficient coding emerges along the visual hierarchy. Neuropixels recordings across the rat ventral-like visual hierarchy (V1, LM, LI, LL), using maximum-entropy textures in which 1-, 2-, 3- or 4-point correlations can be varied independently, showed that 1-point correlations were robustly decodable throughout the hierarchy, whereas higher-order correlations showed pronounced area dependence: both LI and LL displayed the same 2-point > 4-point > 3-point ordering observed in natural-image variability and perceptual sensitivity, an ordering absent from earlier visual areas. Generalization tests, withholding entire texture instances from training, showed a performance gap between standard and instance-held-out decoding that progressively decreased along the hierarchy, indicating increasing invariance to texture instance. These results identify a population-level cortical signature consistent with efficient coding of natural image statistics that emerges at late stages of the visual hierarchy.
Luigi Valio — Suor Orsola Benincasa University, Naples
Seeing tools together: a perceptual and behavioural signature of the cooperative mind
Human technology developed within shared social contexts, and convergent evidence suggests that cooperative settings served as a crucible for cumulative technological culture — though it remains unclear whether this social dimension leaves a signature on perceptual mechanisms for tool processing. This study asked whether the mere co-presence of another person alters how the human visual system explores technological artefacts. In Experiment 1, 22 participants freely viewed isolated tools either alone or in the presence of an unfamiliar peer while eye movements were recorded; growth-curve analyses showed that co-presence altered the temporal balance between functional and manipulation-related regions, with gaze initially favouring functional information in both conditions but shifting toward manipulation-related information more strongly in the shared condition. Experiment 2a extended this to tool–object pairs in 30 participants: co-presence selectively increased gaze allocation to tools, while object-directed gaze remained comparable across conditions. Experiment 2b tested behavioural consequences in 53 participants, who responded faster in the shared than solo condition in both a usability-judgement and a recognition task, with no evidence the benefit's magnitude differed between tasks. Together, these findings show that mere social co-presence can reorganize visual exploration of tools and facilitate subsequent task performance, even without explicit interaction or shared goals — woven into the perceptual dynamics of tool processing that shaped human technological cognition.
Mario Dalmaso — University of Padova
Culture shapes the spatial representation of speed and time
Numerical and non-numerical magnitudes are often mapped onto horizontal space, and cultural habits such as reading and writing direction are known to influence these spatial mappings, though evidence to date remains limited and mixed. This study investigated whether culture modulates the spatial representation of two non-numerical dimensions closely related to reading and writing habits — the visually perceived speed of random dot kinematograms, and perceived face age — testing whether reading and writing direction shapes their horizontal spatial mapping, and whether both dimensions show distance effects supporting their representation as ordered magnitudes. Italian participants, who read and write left to right, and Iranian participants, who read and write right to left, judged in one study whether a centrally presented random dot kinematogram was slower or faster than a reference speed, and in a second study categorised a centrally presented face as younger or older than a reference face, responding with left and right keys. Opposite spatial mappings emerged across the two cultural groups for both dimensions: Italian participants showed a left-to-right mapping of visual speed and face age, while Iranian participants showed the reverse pattern; a reliable distance effect, with faster responses as the difference between target and reference increased, emerged in both studies and both groups. These converging effects for visual speed and face age support the view that spatial mappings of these dimensions are linked to cultural experience.
Elisabetta De Simone — SISSA, Trieste
Perceptual processing differences and their effect on cognitive mechanisms of reading
How do differences in perception affect cognition more generally? This work addressed the question through a scoping review and new experimental evidence on the differences and similarities between print reading in sighted individuals and braille reading in visually impaired individuals. A scoping review of behavioural studies on word recognition (1997–2026), screening 1,157 records down to 59 articles across adults and children, found evidence in braille reading for benchmark effects such as word frequency and length, though some item characteristics affected braille readers more than print readers — for example, both reading fluency and accuracy were affected by increasing word length in braille readers but not in print readers — and braille reading outcomes were also influenced by tactile-processing factors such as degree of visual impairment, finger sensitivity, and one- versus two-handed reading patterns. Since no studies had directly compared eye-movement and finger-movement measures, a finger-tracking prototype — a tablet with a superimposed braille sheet, back-ended by the authors' own ReadLet infrastructure — was devised to administer reading tasks: 38 braille readers read two texts (silently and aloud) and answered comprehension questions, with finger movements and related measures captured, while eye-tracking measures were captured from 28 print readers on the same materials. Preliminary analyses confirmed the reviewed literature: word-length effects appeared only in reading aloud for print readers but also in silent reading for braille readers, while word frequency affected both groups similarly, suggesting reliance on comparable predictive strategies. Differences in perceptual processing thus appear to affect benchmark reading effects only to a limited extent, suggesting similar underlying cognitive mechanisms in printed and braille reading, even as tactile-processing specificities shape braille reading outcomes.
Massimo Grassi — University of Padova
From Carl Stumpf to early auditory processing: tonal fusion, consonance, and musical aesthetics
Since Carl Stumpf, tonal fusion — the tendency for simultaneous tones to be perceived as a single auditory object — has been proposed as a perceptual basis of musical consonance, though its relation to consonance remains debated because aesthetic judgments may not simply mirror perceptual fusion. Two laboratory experiments, each testing around 100 participants with synthetic complex tones, had participants complete a speeded one-sound/two-sound discrimination task followed by unspeeded ratings of fusion and consonance — Experiment 1 on a targeted set of theoretically central intervals, Experiment 2 extending to a chromatic set from unison to major ninth. Across experiments, speeded responses and explicit fusion ratings revealed a highly structured but compressed fusion profile: a small set of intervals — especially the octave, fifth and fourth, and to a lesser extent the major third — stood out as relatively fused, while most other intervals formed a more weakly differentiated low-fusion region; reaction times were longest for intervals with intermediate one-sound probabilities, suggesting decisional ambiguity rather than fusion strength alone. The behavioural fusion profiles correlated strongly with the predictions of a periodicity-based model that formalizes tonal fusion as an autocorrelation-like property of the auditory periphery's temporal response, and consonance ratings were also strongly related to fusion, though the correspondence was not complete. Rather than lying along a smooth continuum, musical interval perception appears organized around a few perceptually privileged intervals embedded within a broader field of weakly differentiated alternatives — supporting Stumpf's view of tonal fusion as a robust perceptual dimension of musical intervals, while showing that aesthetic consonance builds on, but does not simply reproduce, auditory-object organization.
Stefano Uccelli — University of Milano-Bicocca
Faster, yet rather precise: joint synchronization strongly biases timing accuracy with a minor cost to precision
When people synchronize their movements during joint actions, they typically drift into a systematic acceleration of the prescribed tempo — the "joint rushing" phenomenon, documented in pairs, larger groups, and naturalistic rhythmic interactions. While prior work examined how social interaction biases timing accuracy (period correction), whether it also affects timing precision (phase correction) remained largely unaddressed. Pooling 110 participants across three independent finger-tapping experiments, this study analysed the Weber fraction — the standard deviation of tapping errors relative to interval duration — comparing individual and social performance across sub-second intervals: participants listened to an interval duration (300–700 ms) and then maintained it by tapping with their index finger, either alone or together with a partner whose hand they could see. Joint synchronization elicited a dramatic, duration-scaled acceleration in accuracy but imposed only a modest, roughly duration-independent loss in precision, preserving its reversed-J function; rushing was about three times steeper in the joint than in the solo condition, while Weber fractions were only slightly higher. This indicates that social synchronization primarily alters the period-correction mechanism while largely sparing phase correction — a distinction relevant to current models of temporal coordination in social behaviour, and a possible explanation for why people moving together feel in sync even as they gradually speed up and drift away from the prescribed tempo.
Laura van Zantwijk — Humboldt University of Berlin
Does the visual adaptation of causality spread across objects?
Observers report a vivid impression of "launching" when one object approaches another, which then moves in the same direction upon contact; this perception of causality can be altered by visual adaptation, since after prolonged exposure to launch events observers become less likely to report a launch in subsequent ambiguous test events ranging from launching to passing. Three experiments asked whether this adaptation is spatially restricted to the point of contact or can spread to areas that made no direct contact between the two objects. In Experiment 1 (n = 12), a single disc collided with a vertical stack of five discs, always at the bottom of the stack, and test events were shown at the centre, top or bottom of the larger object; adaptation effects appeared only for test events at the bottom, indicating strong spatial specificity with no spread of causality. In Experiment 2 (n = 12), a larger moving rectangle collided with a single disc; this time adaptation showed a small spread across the launching rectangle, with effects at the centre and top but not the bottom, while a motion-offset control stimulus produced no adaptation. In Experiment 3 (n = 15), a larger rectangle launching another larger rectangle produced no adaptation, whereas a control disc launching another disc produced strong adaptation at all three locations, and the Experiment 2 pattern was replicated. Together, these findings suggest that perceived causality relies on spatially precise mechanisms tied to the point of contact, but that adaptation can also spread to parts of an object that had no direct contact with the launched object.
Nicola Bruno — University of Parma
Lateralization of spatial attention: Evidence from line bisections by football officials
In the healthy brain, hemispheric lateralization biases attention to the left hemifield — a phenomenon known as pseudoneglect — and it is currently unclear whether pseudoneglect can be modulated by training. Football officials provide a natural experiment to address this question, since referees deploy attention across the field in all directions, whereas assistant referees monitor the guidelines back and forth and allocate attention mostly to the left. Attentional biases from referees and assistants were assessed using a standard line bisection task. Results documented individual differences in bisection errors, with a prevalence of errors signalling left attentional biases (pseudoneglect), as expected; critically, this left bias remained essentially constant for the assistants group but decreased as a function of months of expertise for the referees group. At least in this sample, prolonged training in allocating attention to both visual hemifields appears to counteract the left attentional bias typical of most individuals.
Valter Prpic — e-Campus University
Joint Processing of Arrow Direction and Number Magnitude Strengthens Spatial-Numerical Associations: Evidence from a Modified Flanker Task
Whether numbers are inherently represented in a spatial format is a long-debated question in numerical cognition. Spatial-Numerical Associations (SNAs) — the tendency to associate small numbers with left/bottom space and large numbers with right/top space — have traditionally been studied with bimanual response settings, but recent unimanual paradigms, such as a modified Implicit Association Test combining numbers and arrows in sequential trials, found SNAs only when spatial and magnitude codes were explicitly contrasted, questioning that inherent link. To investigate this further, a modified flanker task presented central target digits surrounded by a matrix of flanker arrows simultaneously, with horizontal and vertical arrows evaluated across magnitude classification and parity judgment tasks. When arrows were task-irrelevant, no significant SNA or MARC (linguistic markedness of response codes) effect emerged for either dimension; but when participants judged arrow direction and numerical features conjunctively via an AND-rule, the paradigm revealed unusually large SNAs in magnitude classification and robust MARC effects in parity judgment. These results offer no support for an inherent spatial-numerical link, suggesting instead that robust SNAs require explicit spatial codes combined with semantic number processing — and the large, well-behaved effect sizes obtained suggest the modified flanker task is a promising paradigm for studying individual differences and effect modulations, helping resolve mixed findings from earlier bimanual and unimanual approaches.
Ian M. Thornton — University of Malta
Are we underestimating afterimages? Simulating dynamic illusions in search of an answer
Whenever the eyes remain still while objects move in the world, adapted contrast responses are released at points of occlusion — commonly known as afterimages. Depending on fixation duration, scene articulation and lighting conditions, observers may be more or less aware of such afterimages, but they are nonetheless ubiquitous during everyday viewing; aside from a few classic demonstrations, vision science generally assumes afterimages are a form of early "noise" that is filtered out during upstream processing. This talk presents situations in which afterimages instead appear to contribute as "signal" rather than "noise" during perceptual experience, beginning with a general introduction to dynamic illusions that involve misperceiving an object's true path of motion, then focusing on two such illusions that seem to involve afterimages: Partial Object Doubling (Thornton & Riga, 2024) and the Mainz–Linz Illusion (Thornton, Riga, Zdravković & Todorović, 2025). For the latter, a new simulation framework modelling both motion-energy and contrast-adaptation responses shows that either channel could, in theory, account for the path deviations observed in human viewers.
Poster session — Foyer
26 posters. Equipment presentations continue in parallel (ULTRA in Room 1_C, NIRx/Artinis in Room T_A).
Venue: Building F2, foyer of Room T_B, ground floor, San Giovanni campus, Via Edoardo Weiss 9, Trieste · map ↗
Final programme.
History
The Trieste Symposium on Perception and Cognition began in 1994 as a meeting point for researchers of visual perception, attention and cognitive processes, named in honour of Gaetano Kanizsa — read more on the Kanizsa Lecture homepage.
The symposium has repeatedly travelled to partner universities: the 2021 edition, for instance, was hosted by the Department of General Psychology at the University of Padua, while the 32nd Kanizsa Lecture (2024) was held at the University of Milano-Bicocca by the Department of Psychology in collaboration with the Society for Gestalt Theory and its Applications (GTA).
Organizing committee of the 2026 edition
Archive
Editions prior to 2010, dating back to 1994, are not yet linked here.
Contact
For programme information, to propose a contribution, or to join the mailing list, write to trieste.symposium@gmail.com.