• Huntington's Disease
• language learning
• temporal attention

Today our knowledge of the neurobiological bases of language is still very scarce if compared to the one of vision, memory and motor functions. This is mainly due to a lack of animal models and to the extreme complexity of language. Our ability to communicate through a structured and meaningful ensemble of sounds results from an incredibly huge variety of cognitive processes and from the interaction of different brain areas. Surely the neural substrates of language are not exclusive for it, but also subserve other functions. Most of them were initially responsible of more basic behaviours and were successively recycled once linguistic abilities started to appear during evolution (Ullman 2016). In this context, if we want to understand how language processing works, we should study first the mechanisms which underpin these higher order skills, such as attention, time perception, executive functions and procedural memory and the brain areas connected to them. Between these brain areas the striatum appears to be particularly interesting, since it’s involved in all the aforementioned behaviours. However up to now very little is known about striatal contribution to language learning. A way to approach this topic is the study of neurodegenerative diseases connected to striatal degeneration, like Parkinson’s or Huntington’s disease (HD), to see possible alterations in the behavioural response. The aim of the present study was to analyze temporal orienting abilities and language learning of non-adjacency rules in a group of presymptomatics and early stage Huntington disease gene-carriers and controls matched for age, gender and education. The study included a behavioural and an imaging part. For what concerns the behavioural part two different experiments were conducted. The first experiment tested temporal orienting abilities trough a speedy detection task. Participants were asked to press a key as soon as possible as they heard a target sound. An auditory cue (valid in 75% of times) allowed participants to orient attention to the precise time point of target sound appearance. In this task we collected reaction times, which depended on both the effect of exogenous temporal expectation (foreperiod effect, probabilistic information linked to the passage of time) and of endogenous temporal expectation (prediction of the target appearance allowed by the auditory cues). The second experiment tested the participant’s ability to learn non-adjacency rules (NAD) in an artificial language. NAD are associations between words which are separated by intervening elements that are statistically independent from them. They can be represented through an AXC structure, were the X identifies one or more intervening elements. This kind of structures are common in natural languages, where, for example, we can find dependencies between auxiliaries and inflectional morphemes (e.g. is screaming, has broken), as well as dependencies for number agreement (e.g. the dogs…. are barking). Here temporal orienting is hypothesized to play a role since the first element in the dependency (e.g. is) can be used as a temporal cue to predict the later appearance of the second element of the dependency (e.g. –ing). The experiment was subdivided in 2 parts: a learning phase and a test phase. For the learning phase, words were combined to form rule phrases (AXC) and filler (XXX) phrases. Following the structure used in previous studies (Gómez 2002; Gómez and Maye 2005; López-Barroso et al. 2016), rule phrases were AXC phrases (e.g. tagi-male-sira, tagi-fuse-sira), where the initial word (A) determined the third (C) regardless of the middle element (X); three different AXC rules were used. In order to manipulate the amount of attention given to the rules a target word had to be detected. The target was the last word of one of the rules (target condition) whereas the other two rules did not contain the target word, allowing the subjects to ignore them in principle (non-target conditions). The test phase included a test for implicit and a test for explicit learning. In the implicit test if participants learned the specific dependencies in AXC structures they should respond faster in rule phrases than in non-rule, since rule presence allowed for prediction. Immediately after the rule learning task an explicit learning test was performed to see if attention (target presence) affected explicit learning. Participants were asked to discriminate phrases following the rules to which they had been previously exposed from those violating these rules. We wanted to see if HD had an impairment in NAD learning and if this impairment eventually correlated with the performance in the temporal attention task. This correlation could possibly be due to the fact that both behaviours depend on the same neural substrates, supposedly the ones which degenerate in HD. To assess this, we first calculated the correlation coefficients between different attention and language results, then we performed a voxel-based morphometry analysis on T1 images taken from the HD group. In this latter analysis we focused only on the brain areas that showed a shrinkage (supposedly due to degeneration) in HD with respect to controls. By applying a general linear model, we looked for correlations between the voxel intensities and the behavioral parameters taken from the language and attention experiments.