Purpose & Mission
The Pittsburgh Science of Learning Center: Studying Robust Learning with Learning Experiments in Real Classrooms
Statement of Purpose
Leverage cognitive theory and cognitive modeling to identify the instructional conditions that cause robust student learning.
The Pittsburgh Science of Learning Centers (PSLC) two main goals are to enhance scientific understanding of robust learning in educational settings and to create a research facility to support field-based experimentation, data collection and data mining. PSLC is advancing both basic research on learning in knowledge-rich settings and applied research by contributing to a scientific basis for the design and engineering of educational approaches that should have a broad and lasting effect on student achievement.
In many studies of learning and in many educational settings, learning is assessed immediately following instruction using test items like those presented in instruction. In contrast to such immediate learning assessment, we seek methods to produce and measure robust learning, by which we mean learning that is retained for long durations, transfers to novel situations, or aids future learning (Barnett & Ceci, 2002; Bransford & Schwartz, 1999; Singley & Anderson, 1989). In contrast to the education wars that have plagued progress in the learning sciences and in educational practice, we do not pit foundational skill building against sense-making and conceptual understanding, but instead believe we must address both to improve robust learning. These wars continue, in part, because we do not have adequate scientific basis to guide educational decision-making. We need rigorous, sustained scientific research in education, as called for by the National Research Council (Shavelson & Towne, 2002), and a key part of such sustained research is to better unify and integrate the proliferating variety of todays educational and learning science theories. As the saying goes, many theories in the learning sciences are like your toothbrush: everyone has one and no one uses anyone elses.
The center's integration strategy is targeted at supporting and advancing a shared theoretical framework of robust learning across academic content domains, particularly, science, math, and second language learning. The framework has both a micro-level, which builds off the Pittsburgh tradition of detailed, computationally-based theories of cognition and learning (e.g., Anderson & Lebiere, 1998; McClelland, Cleeremans, & Servan-Schreiber, 1990; Newell, 1990), and a macro-level, which involves organizing, sharing and ultimately unifying existing and emerging educational theories and approaches. We are building theoretical links between micro-level computational theories and critical macro-level instructional choices. Following Pasteurs Quadrant (Stokes, 1997), we are pursing both bottom-up applications of micro-level theory and top-down use inspired basic research, whereby questions from the macro-level drive basic research on learning (see Figure 2). Many macro-level learning principles have been proposed (e.g., Bransford, Brown & Cocking, 2000; Clark & Mayer 2003; National Council of Teachers of Mathematics, 2000), but the evidence base for these principles is incomplete. Equally important, the theoretical support for these principles is often limited and does not rest upon a precise, explanatory and predictive theoretical framework the kind of framework computational theories might provide.
To build this framework of causal and generative principles, we need a sustained scientific effort built upon efficient methodologies and facilities for engaging in learning science research of both high internal validity (laboratory quality, principle-testing studies) and high ecological and external validity (in real learning settings and tested across multiple domains). PSLCs facilitation strategy is to create such an efficient research infrastructure, which we call LearnLab. The LearnLab involves research agreements with educational institutions (K12 schools and colleges) to allow researchers regular access to run experiments as part of real, operating courses. These are state-of-the-art technology-enhanced courses that are highly instrumented for fine-grain long duration recording of student learning activities. The LearnLab provides researchers with access to courses in which to run rigorous studies in real field settings. These in vivo learning experiments have three key features: 1) internal validity: students are randomly assigned to either a treatment or control condition and a single variable distinguishes the treatment from the control, 2) ecological validity: the experimental setting is in the field with real students pursuing real course goals over realistic durations and the control condition represents existing practices, and 3) the variable distinguishing the treatment and control is a more fine grained instructional method or principle, not a coarse aggregate variable like a new textbook, teacher professional development program, or technology. Pure laboratory experimentation involves 1 and 3, but not 2. Design experiments (Brown, 1992; Cobb, Confrey, diSessa, Lehrer, & Schauble, 2003; Dede, 2005) involve 2, but not 1 and 3. Randomized field trials involve 1 and 2, but not 3. In vivo learning experimentation pre-dates PSLC (e.g., Aleven & Koedinger, 2002; Brown & Palincsar, 1989; Howe, Tolmie, & Rodgers, 1992; McNamara & Scott, 1999; Evens, Spitkovsky, Boyle, Michael & Rovick, 1993; Leonard, Dufresne & Mestre, 1996; Metcalfe, 2006). LearnLab helps make this method more feasible for researchers to use and more visible to the research field. PSLCs LearnLab is unique in 1) providing learning researchers an open facility for in vivo experimentation, 2) systematically collecting longitudinal moment-by-moment student learning data in multiple courses running and multiple sites, and 3) making such data freely available. Although rigorous learning experiments can be done at larger grain sizes (e.g., policy studies) and smaller grain sizes (e.g., cognitive neuroscience), in vivo experimentation in the LearnLab is ideal for addressing questions of robust learning because LearnLab courses provide built-in tests of long-term retention, transfer and preparation for future learning.
PSLCs research focus is to make students more effective learners while discovering the underlying principles that can be used across domains. As illustrated in Figure 2, our research and quest for understanding (e.g., research cluster activity) drives better applications and our consideration of use (e.g., course committee activity and in vivo experimentation) drives new basic research questions.
Figure 3 illustrates the interlocking research and development activities of PSLC and, in particular, how shared theory development is supported by the LearnLab facility, including in vivo studies run in LearnLab courses and enabling technologies to support all aspects of research. A major goal is to minimize the demanding set up costs usually required of individual researchers who want to perform in vivo studies in schools (see Figure 4).
LearnLab facilitates in vivo experimentation by researchers from around the world by providing social and technological facilities for creating, running, and analyzing such experiments more easily than ever before. PSLC provides and instruments six courses: Algebra, Geometry, Physics, Chemistry, Chinese and English as a second language. These courses are existing courses in use at a number of high schools and college sites that have agreed to host in vivo experimentation. Our initial goal was that LearnLab would grow in 5 years to support at least 42 in vivo learning experiments a year. Already in our fourth year, we host an average of more than 30 in vivo learning experiments per year. We also support some studies and provide data from several affiliate courses.
In addition to the core research strategies of integration and facilitation, our third and fourth strategies are to enhance diversity and education in the learning sciences and in the beneficiaries of learning science. LearnLab both requires and affords new training of learning scientists to take advantage of this new paradigm. LearnLab enriches the scientific infrastructure by establishing extensive collaborations among a new cadre of current and future cognitive, developmental, and educational psychologists, statisticians, psychometricians, computer scientists, instructional designers, and discipline-based education researchers (e.g., Chemistry or Chinese education researchers). In our first five years, we plan to provide 50-100 pre-doctoral and post-doctoral students research experience in developing and using LearnLab to general robust learning principles. Already we have provided research experience to more than 65 pre and post-doctoral students. This does not include summer interns (another 48) nor does it include undergraduates who have worked in the center (approximately another 25 per year). Experimental manipulations that demonstrate substantial learning gains have been and will continue to be incorporated in LearnLab courses and schools and eventually will reach many other schools.
Diversity efforts focus on activities to increase the pipeline of racially diverse students and researchers in the learning sciences from the high school level to center leaders. Our education and diversity strategies are necessary to enable the new learning science research paradigm that LearnLab will make possible.