In 1997 Elsbeth Stern was hired as a senior research scientist to establish a research project on science and mathematics learning for elementary school children, called ENTERPRISE (Enhancing KNowledge Transfer and Efficient Reasoning by Practicing Representation In Science Education).
The project was based on a domain-specific perspective of cognitive development, arguing that the difficulties experienced by younger children in coping with specific cognitive demands can be traced back to their lack of domain specific conceptual and procedural knowledge. In this perspective, learning environments for young children should focus on laying the foundations for the kind of knowledge acquisition that, in the upper grades of school, is most relevant and challenging. Knowledge dealt with in all kinds of academic contexts is based on symbolic systems, such as script, formal mathematical language, pictures, and diagrammatic representations.
Symbols can be understood as mental tools that support the construction of meaning in concepts, ideas, or plans. The ENTERPRISE project focused on fostering elementary school students’ understanding of science concepts by providing students with reasoning tools such as visual representations, informal notations or other symbolic systems. It was shown that with the help of cognitively structuring tools like balance beams or graphs, already third graders were able develop a deeper and transferable understanding of scientific concepts. Specifically, in the PhD theses of Susanne Koerber and Anja Felbrich these effects of using visual representations such as the Cartesian graph on students’ conceptual understanding were investigated. In a cooperation project with Kornelia Möller from the Department of Elementary Science Education of the University of Münster, Ilonca Hardy and Elsbeth Stern investigated the short-term as well as long-term effects of cognitive support such as use of cognitively structuring teacher statements on elementary school students’ conceptual understanding within the context of “floating and sinking.” In several studies, they found evidence for long-term conceptual enhancement through the use of reasoning tools such as balance beams or Cartesian graphs within complex science learning environments in elementary school (Hardy, Jonen, Möller, & Stern, 2004; 2006; Hardy, Schneider, Jonen, Möller, & Stern, 2005) .
The effects of cognitive tool use on the emergence of conceptual knowledge was addressed in two dissertations finished in 2006. Michael Schneider investigated how practicing with number line games affects the relationship between procedural and conceptual knowledge (Schneider, 2006). Henrik Saalbach ran cross-cultural studies in order to explore the impact of grammar as a cognitive tool on conceptual development (Saalbach & Imai, 2006; Saalbach & Imai, in press).

Cognitive Tool Use and its Effects on the Interaction between Conceptual and Procedural Knowledge

Conceptual knowledge, as seen by cognitive theories of learning, provides an in-depth understanding of the principles of, and interrelations among, pieces of knowledge in a domain. Procedural knowledge allows for the efficient and goal-directed solution of routine problems and is represented, on the cognitive level, by production rules with different degrees of automaticity.
In the literature addressing the interrelations between conceptual and procedural knowledge, three basic approaches can be distinguished: concepts-first approaches, which are based on the assumption that the acquisition of conceptual knowledge takes place, as a rule, prior to that of procedural knowledge and, subsequently, causes an increase in procedural knowledge; procedures-first approaches, which postulate the exact opposite with respect to the sequence of acquisition and the direction of causation; and the Iterative Model, which holds that the sequence of acquisition depends on experience and that causal relations are bi-directional. In his thesis, Michael Schneider took up the challenge to find measures allowing for the separation of conceptual and procedural knowledge which, in actual behavior, are inextricably interwoven.
Predictions based on the Iterative Model were tested in the well researched field of decimal fractions against predictions based on the concepts-first and the procedures-first approaches. Four measures of conceptual knowledge (evaluation of procedures, translation into diagrams, size comparison, and written explanations) and four measures of procedural knowledge (problem solving accuracy, problem solving duration, asymmetry of access, dual-task costs) were used, all of them adapted from published studies.
In a first study, knowledge types were modeled in structural equation models (SEM) as two latent factors underlying the eight manifest measures, thus permitting predictive relations between factor pairs to be examined across three points of measurement in a cross-lagged panel design. The sample consisted of 204 fifth-graders and sixth-graders.
The eight measures were shown to possess high convergent validities, but low divergent validities. Inter-correlations between the latent factors were very high. This proved to be an obstacle to the analysis of predictive relations, which provided some clues, but no unambiguous evidence for the validity of the Iterative Model.
In a consecutive study with 84 fifth-graders, it was investigated whether, nonetheless, procedural and conceptual knowledge can be fostered independently of each other by practicing with cognitive tools either in a more procedural or in a more conceptual manner. Two learning environments based on the "catch-the-monster game" were realized. In a 2x3 experimental design with two independent treatment groups and three measurement points, no between-group differences were revealed. All in all, results show that conceptual and procedural knowledge are much more strongly interrelated than one would expect from all previous studies which, using only manifest measures, were more prone to measurement errors. Therefore, the Iterative Model could be neither validated nor invalidated. Although the distinction between both kinds of knowledge remains useful from a heuristic point of view, cognitive competence in a subject area like mathematics seem to rely on the integration of well developed procedural as well as conceptual knowledge.

The Impact of Grammatical Language Tools on Conceptual Structures

From the very beginning, humans structure their complex environment by grouping stimuli according to their similarity by various criteria. In the case of concrete objects, shape is a very salient feature strongly calling for categorization. Another common way of classifying objects uses thematic relations which can be based on co-occurrence or interaction in space or time, or functional or causal relationships (e.g., carrot/rabbit, scissors/paper, cow/milk). Still another basis of conceptual structuring already found in younger children concerns non-obvious taxonomic relations. Taxonomic categories are differentiated into levels of varying specificity (e.g., animal, dog, collie) related by class inclusion (e.g., a collie is a dog, a dog is an animal, a Collie is an animal).
Three kinds of tasks were used to find out whether objects are grouped on the basis if their form, or of their thematic or taxonomic relations, respectively:

  1. In the non-lexical classification task, participants were asked to select the item that best matched the standard object.
  2. In the word-learning task, participants were asked to extend a novel label that was given to the standard ("This is called FEP. What else is called FEP?")
  3. In the property induction task, participants were taught a novel non-perceptual property of the standard object (e.g., "Look, this one has IDOFORM inside.") and asked to select the item that would be most likely to have the same property.


According to the widely accepted linguistic relativity approach, the conceptual structure of children and of adults can be affected by the grammar of their mother tongue. Cross-language studies have revealed that certain grammar elements can affect cognitive development mainly by influencing the foci of children's habitual attention, thus leading them to engage in certain processes of categorization and conceptualization. A particularly strong impact of grammatical structures such as count/mass syntax or gender grammar has been shown by several studies. The grammatical structure that was the focus of Henrik Saalbach's research was the Chinese system of numeral classifiers by which nouns are grouped into linguistic categories just as they are by the system of grammatical gender in German. Unlike gender systems, however, there are more than 20 classifier categories in Chinese that are actually used, organized around semantic features such as shape, function or animacy. For example, tiao, a common Chinese classifier for long and flexible things, includes many things from different taxonomic categories - fish, dogs, rivers, roads, pants, and more - and even crosses the animate-inanimate boundary.
The core question was whether the classifier system affects children's concept in any significant ways. According to the semantic nature of classifier categories (i.e., the fact that shape is an important semantic feature), it was hypothesized that both Chinese children and adults may have a stronger tendency than speakers of non-classifier languages to organize objects on the basis of their shape rather than on taxonomic or thematic relations.
German and Chinese speakers of three age groups (3-year-olds, 5-year-olds, adults) were tested on a match-to-standard generalization paradigm in the three mentioned formats: non-lexical classification, word learning and property inference. In all task contexts, a child was shown a picture of the standard object (e.g., banana), and was asked to choose one item out of three choice items: a taxonomic item from the same super-ordinate category (grape), a shape/classifier item (feather), and a thematic item (monkey). All objects were represented by couloured line drawing.
Differences between German and Chinese Participants were only found in the non-lexical classification task. Chinese speakers, especially children, pay more attention to shape similarity than German speakers, given that shape is a prominent semantic feature in classifier categories. In the property inference task, the majority of participants referred to taxonomy relations. In the word learning task, Chinese as well as German children judged on the basis of shape while all the adults considerend taxonomy relations.
In sum, this research demonstrates that children are extremely flexible and adaptive learners who make use of various environmental cues which may help them to structure the complexity of the world. Grammatical language tools are among such cues and are particularly used in situations not constrained by any more salient information. The results suggest that deliberate language input is important for young children's environment. Long before the youngsters can actively use the correct grammar rules in language production, they make use of them when complex incoming information has to be handled to form useful conceptual structures.

Further activities of the ENTERPRISE group and continuation at the ETH Zürich

Since 2004, Elsbeth Stern has become increasingly involved in discussions on whether neuroscience can inform educational reform (Stern, 2005a, b). On the one hand, she felt responsible for scaling down unrealistic expectations about the educational implications of neuroscience and emphasizing the importance if more traditional empirical educational research. On the other hand, she acknowledged that collaboration between neuroscience and educational research provides unique opportunities to gain insight into the interaction between mind and brain in the cultural context. In 2005, Elsbeth Stern and Michael Schneider headed and coordinated the research program NIL (Neuroscience, Instruction, and Learning) which is funded by the German Ministry of Education and Science. Meanwhile, NIL provides funding for six research projects in which neuroscientists and educational researchers address questions of brain functioning during school-related activities.
In 2006, the ENTERPRISE group moved to the Swiss Federal Institute of Technology Zurich (ETH Zürich), where Elsbeth Stern was appointed Professor for Research on Learning and Instruction and where she heads the masters program for teacher students in secondary higher education. Despite this shift toward research in learning and instruction in the upper grades, the theoretical focus of the ENTERPRISE project has been maintained. The major concern will be to explore how, in the field of science learning with cognitive tools may optimize the way incoming information is connected with knowledge that already exists in the mind. In continuation of the research on the interaction between intelligence and knowledge (Grabner et al., 2006), special emphasis will be placed on the question of how individual differences in cognitive resources affect learning in the area of science and engineering.

Key References

Grabner, R., Neubauer, A., & Stern, E. (2006). Superior Performance and Neural Efficiency. Brain Research Bulletin, 69, 422-439.

Hardy, I., Jonen, A., Möller, K. & Stern, E. (2006). Effects of Instructional Support Within Constructivist Learning Environments for Elementary School Students’ Understanding of “Floating and Sinking.” Journal of Educational Psychology, 98(2), 307-326.

Hardy, I., Schneider, M., Jonen, A., Möller, K., & Stern, E. (2005). Fostering Diagrammatic Reasoning in Science Education. Swiss Journal of Psychology, 64(3), 207-217.

Hardy, I., Jonen, A., Möller, K., & Stern, E. (2004). Die Integration von Repräsentationsformen in den Sachunterricht der Grundschule. In J. Doll & M. Prenzel (Hrsg.), Bildungsqualität von Schule: Lehrerprofessionalisierung, Unterrichtsentwicklung und Schülerförderung als Strategien der Qualitätsverbesserung (S. 267-283). Münster: Waxmann.

Saalbach, H., & Imai, M. (in press). The scope of linguistic influence: Does a classifier system alter object concepts? Journal of Experimental Psychology: General.

Saalbach, H., & Imai, M. (2006). Categorization, Label Extension, and Inductive Reasoning in Chinese and German Preschoolers: Influence of a Classifier System and Universal Cognitive Constraints. Proceedings of the 27th Annual Conference of the Cognitive Science Society.

Saalbach, H., & Stern, E. (2004). Differences between Chinese morpho-syllabic and German alphabetic readers in the Stroop interference effect. Psychonomic Bulletin & Review, 11, 709-715
Schneider, M. (2006). Konzeptuelles und prozedurales Wissen als latente Variablen: Ihre Interaktion beim Lernen mit Dezimalbrüchen. Unpublished PhD thesis, Technische Universität Berlin, Berlin.

Stern, E. (2005a). Pedagogy meets Neuroscience. Science, 310, 745.

Stern, E. (2005b). Brain goes to school. Trends in Cognitive Science, 9, 563-565


Elsbeth Stern
Ilonca Hardy

Predoctoral Research Fellows:
Anja Felbrich
Michael Schneider

Cooperation Partner:
Prof. Dr. Kornelia Möller, Universität Münster
Dr. Susanne Koerber, Universität München
Dipl.-Psych. Henrik Saalbach, University of Beijing