Browsing by Subject "neurotieteelliset tietokannat"

In brain imaging research, two types of inferences are regularly employed: direct inference proceeds from a cognitive function to observed brain activation, whereas inverse inference takes an opposite direction from observed activation to a cognitive function. Inverse inference is more problematic of these two. Its hasty application could lead to errors in the interpretation of experimental results. In this study, conditions for the succesful use of inverse inference are analyzed. To start with, inverse inference should always be part of the testing process of a neurocognitive model, and there should be at least preliminary understanding of the neural correlates of cognitive functions. With the help of Bayesian analysis, quantitative estimates for the strength of inverse inference could be calculated. The bayesian analysis shows that this strength is inversely proportional to the number of different experimental tasks associated with activations in the region of interest. There is also an important distinction between two types of inverse inference: constrained inference is restricted to the activations associated with one type of experimental task, wheras unconstrained inference involves all the tasks that activate the region of interest. Especially unconstrained inference is strengthened by the use of large pools of experimental results, which are recently collected to the neuroscientific databases such as BrainMap or NeuroSynth. One problem for inverse inference is that there are not any commonly accepted taxonomies of cognitive functions. The lack of unitary vocabulary could leave the mapping between mental functions and brain mechanisms indeterminate. As a solution, some researchers have developed cognitive ontologies, which are formal knowledge structures with an explicit goal to define unitary empirically-based cognitive concepts. As a conclusion, it could be stated that if the above mentioned considerations are carefully taken into account, the employment of inverse inference becomes acceptable in the context of brain imaging studies.