False estimated is forearmed - model describes how information affects our perception
When we estimate something, we subconsciously use recent experiences. Researchers at the Ludwig Maximilians University (LMU) Munich and the Bernstein Center Munich asked test subjects to estimate distances in a virtual environment. Their results tended towards the mean value of all routes run up to that point. For the first time, the scientists were able to predict the experimental results very well using a mathematical model. It combines two well-known laws of psychophysics with the help of a proposition from probability theory. The study could thus be of fundamental importance for perception research. (Journal of Neuroscience, November 23, 2011)
Why do we estimate the same distance one time long and the other short? The decisive factor is which distances we covered directly before. What may sound trivial, provides important information about how the brain processes stimuli of different strengths and even abstract elements such as numbers. This is what Dr. Stefan Glasauer (LMU), project manager at the Bernstein Center Munich, and his doctoral student Frederike Petzschner experimentally and theoretically. They had test subjects cover distances in a virtual room and then reproduce them there as precisely as possible. As in previous studies, the results were always shifted from the correct value to the mean value of the previously run distances.
The scientists can now provide a general explanation for this phenomenon. Using a mathematical model, they can calculate how previous stimuli affect the current estimate. "Most likely influence of prior experience follows a general principle and may also be true for the estimation of quantities or volumes," says Glasauer. Subjects who were strongly influenced by previous experience in distance estimation, also placed greater weight in angle estimation on their previous experience. In both cases, they learned even without the success or failure of their power to know. learning methods, however, presuppose such feedback.
Until now disputed whether a fundamental principle determines the perception of stimulus strengths such as volume, brightness, or distances. Two important laws of psychophysics, thereby contradicting: the published 150 years ago Weber-Fechner law and the 50 years old Stevens' power. The Munich scientists have now shown that it is possible the two laws bring at least in certain cases very well reconciled.
For this is the Weber-Fechner law with the probabilistic Bayes 'Theorem (1763), which allows evidence to be weighted, combined and converted into the Stevens' power. "We were able to contribute to the solution of a problem that perception researcher busy for over 50 years," Glasauer is therefore convinced. Next, the researchers plan to analyze and clarify whether the model also applies different stimulus modalities such as volume and brightness historical data.
The Bernstein Center Munich is part of the National Bernstein Network Computational Neuroscience (NNCN). The NNCN was founded by the BMBF with the objective of pooling the capacities in the new scientific discipline of computational neuroscience, to network and develop. The network is named after the German physiologist Julius Bernstein (1835-1917).
Original work:
Petzschner F, Glasauer S (2011): Iterative Bayesian estimation as to explanation for range and regression effects - A study on human path integration. J Neurosci 2011, 31 (47): 17220-17229
Source: Munich [LMU]
