Nowadays, students are frequently exposed to different and multiple sources of information, from which they may be asked to perform a variety of tasks such as writing essays, answering comprehension questions or locating specific units of information. Although a significant amount of research has been conducted on how students integrate information from multiple historical texts (ie., Wineburg, 1991; Perfetti, Britt & Georgi, 1995; Rouet, Britt, Mason & Perfetti,1996; Rouet, Favart, Britt & Perfetti, 1997; Britt & Aglinskas, 2002), little has been studied so far on the mental processes involved in integrating information from multiple expository texts, which are the focus of our attention in the present dissertation. On the other hand, similarly to the case of single-text learning with the use of inferential questions (i.e., Vidal-Abarca, Mengual, Sanjose & Rouet 1996; Vidal-Abarca, Gilabert & Rouet, 1998; Rouet et al., 2001), there should be tasks that are more effective than others in making students learn from multiple expository sources at a deep level. Indeed, Wiley and Voss (1999) found that performing argument tasks after reading the multiple sources produced increased conceptual understanding of the main topic presented in multiple texts. Based on the previous empirical background, we formulated the following hypothesis for our study: First, search for a task that induces integration and learning from multiple documents in a greater extent than others. Second, deepen into the mental processes responsible for enabling integration from multiple sources and find behavioral correlates for these processes. And, globally, analyse how students integrate information from multiple expository texts. For this purpose, we selected three texts on bacteria resistance to antibiotics and created two kind of tasks: one global task, similar to a summary, which would make the students actively connect and integrate units of information present in each document; four brief questions, which would only require the attention to one of the single texts to find the answer. We expected that the global task would induce integration and learning from the multiple sources in a greater extent than brief questions, as it would induce a more active processing of the texts. We conducted a first experiment in which 50 university students read the texts and answered the questions, all in a software called Read&Answer, which allows the tracking of the reading and question-answering process. Additionally, half of the sample thought-aloud. We obtained a set of measures related to the reading behavior (i.e., how fast or slow was the reading, if relevant information was read more slowly...) and measures related to the learning outcome (i.e., final learning measures after the experimental session). Results from the first experiment yielded highly interesting results. As we expected, it was the global task the only one inducing long-term learning from the multiple texts, in contrast to brief questions. Moreover, this was because a deep processing of the texts had taken place when answering the global task (i.e., relevant information was read more slowly). Thus, the global task was indeed the most effective in working with multiple documents and, in contrast, brief questions only helped students in constructing a very isolated understanding of the main units of information in the documents. Finally, we obtained a curious result regarding the effects of making students think-aloud: it seemed to overload short-term memory resurces in students, and limited the possibility to acquire long-term learning from the documents. To complement results obtained in experiment 1, we conducted a second experiment in which we included a third kind of task: answering very specific questions (i.e., copy one sentence) which would by no means help understand the main ideas present in the texts, in contrast to brief questions. Results were in agreement with this hypothesis and, interestingly, there was a complete replication in results for the other two kind of tasks. All these results helped us determine that making students answer a global task when using multiple documents is the most effective way to induce integration and long-term learning. __________________________________________________________________________________________________ RESUMEN Presentamos dos estudios en los que nos cuestionamos: (a) qué procesos mentales subyacen a la integración de información a partir de múltiples documentos expositivos; (b) qué tarea sería la más eficaz para promover integración y aprendizaje a largo plazo a partir de documentos múltiples. Para aclarar estas preguntas, realizamos un primer experimento en el que cincuenta estudiantes universitarios leyeron tres textos y resolvieron dos tipos distintos de tareas en ordenador: contestar una pregunta global, que haría que los estudiantes leyeran activamente los textos y conectaran los segmentos relevantes de información en cada uno de ellos; contestar cuatro preguntas breves, que harían que los estudiantes se centraran únicamente en un texto, sin necesitar la conexión entre textos. Esperábamos que la pregunta global promoviera efectivamente integración y aprendizaje a partir de los documentos múltiples. Basándonos en un conjunto de datos on-line de lectura y de resultados de pruebas de aprendizaje final pudimos concluir que la tarea global había hecho que los estudiantes leyeran los textos con más detenimiento, centrándose en la información importante y conectando los segmentos relevantes a través de los textos. Todo ello tenía como consecuencia incrementos en el aprendizaje a largo plazo. En contraste, las preguntas breves hicieron que los estudiantes leyeran más rápidamente y únicamente promovieron una comprensión aislada de las unidades de información. Para complementar los resultados del primer experimento, creamos un tercer tipo de tarea para contrastar en un segundo experimento y que haría que los estudiantes ni siquiera entendieran los segmentos aislados de información. Creamos un conjunto de preguntas específicas que lo único que requerirían para su correcta contestación sería la copia de frases aisladas en los textos. Los resultados de este segundo experimento confirmaron la nula capacidad de las preguntas específicas para aprender a partir de múltiples textos y confirmaron completamente el patrón de resultados para los otros dos tipos de tareas. En conclusión, dos experimentos nos permitieron concluir que hacer que los estudiantes realicen tareas globales a partir de múltiples textos mejora el grado de aprendizaje que pueden obtener de ellos.