ETD

Archivio digitale delle tesi discusse presso l'Università di Pisa

Tesi etd-09052021-113811


Tipo di tesi
Tesi di laurea magistrale
Autore
D'ERRICO, GIOVANNI
URN
etd-09052021-113811
Titolo
Integration in peripheral vision follows optimal rules and may explain crowding
Dipartimento
BIOLOGIA
Corso di studi
NEUROSCIENCE
Relatori
relatore Dott. Cicchini, Guido Marco
Parole chiave
  • optimal behaviour
  • crowding
  • orientation
  • peripheral vision
Data inizio appello
21/09/2021
Consultabilità
Non consultabile
Data di rilascio
21/09/2024
Riassunto
Peripheral vision refers to all visual perception that is not covered by the central part of the retina which has high acuity. It is strongly limited by a reduced cone density and an increased neuronal convergence which lead to a dramatic loss of resolution respect to the fovea. How the cortex processes peripheral information so to yield the illusion of a high-resolution world is still matter of debate. On one side, it seems that peripheral vision can only process a limited number of items, a phenomenon called crowding. On the other side, increasing the quantity of objects in periphery it also increases the confidence of responses. To better understand the transformations operated by cortical neurons, we developed a reproduction task in which the orientation and the reliability of the peripheral target and of the flankers were manipulated in a parametric way. We found a larger integration of contextual information when target was noisier. This result cannot be explained by positional uncertainty however it can be explained by an optimal observer model trying to minimize the overall error. The model can also predict an improvement of representations which explains the higher confidence of the crowding conditions. In a second experiment, we found that maximal integration seems to occur when the overall context created by flankers is equal to the target. On a final note, we measured a reduction of integration while increasing the spacing between target and flankers, like it happens in crowding. Interestingly the model employed here shares many features with those that describe perception in serial dependence. Thus, it is likely describing a general algorithm which handles noisy information, regardless of its nature.
File