INVESTIGADORES
KAMIENKOWSKI Juan Esteban
congresos y reuniones científicas
Título:
Understanding predictability with LMM and cluster-based permutation test: a new approach in multiple comparison issues on LMM
Autor/es:
BIANCHI B; SHALOM DE; KAMIENKOWSKI JE
Reunión:
Congreso; Latin American Brain Mapping Network (LABMAN); 2017
Institución organizadora:
Latin American Brain Mapping Network (LABMAN) - Organization for Human Brain Mapping
Resumen:
As in almost every daily visual task, the brain generates a predictionon the forthcoming stimuli. In reading, this prediction is usuallyoperationalized as the Predictability, the probability of knowing afuture word before reading it. However, this predictions could bebuilt on different factors depending on the stimuli, such as semanticrelations with the context or memory retrieval of known sentences.In the the present study, we aimed to separate these contributionsusing proverbs and common sentences. Classically, EEG data isanalyze by averaging trials within conditions, on a set of chosenelectrodes and time samples, losing information about thedifferences between stimulus and trial execution. This issue wasovercome by designing experiments carefully, keeping the possiblevalues of the variables or conditions low (i.e. increasing the numberof trials to be averaged). Conversely, behavioral experiments ismoving towards more natural and complex scenarios. Particularly, inthe field of reading and eye movements (EM), new statisticalmethods, such as Linear Mixed Models (LMM), allow researchers toexplore the several variables usually involved in these naturalstimuli. However, the EEG poses a new issue: while EM is usuallyone-dimensional measure (fixation time), EEG data ismultidimensional (spatial-temporal), giving rise to a multiplecomparisons problem. The cluster-based permutations test is apowerful and widely accepted method to face this kind of multiplecomparison problem. We combined LMM for each electrode andtime-point and a cluster-based permutation approach to correct formultiple comparisons. Here, we compare two different resamplingmethods with a classical correction criteria (Bonferroni). Using theformer procedures we observed the classical predictability effect(N400), a conspicuous late effect of the word position in sentenceand a sparse effect on the sentence type (proverb vs non-proverb),which is in line with our hypothesis of different components ofpredictability