Higher level students and researchers in biology and medical sciences regularly use the PubMed online search engine (http://www.pubmed.gov), which provides access to the MEDLINE bibliographic database. Despite a lack of formal training to this tool, PubMed has become the gold standard of French neuroscientists for work-related information seeking. Sixteen neuroscience experts of doctorate level or above were asked to perform 5 bibliographic search tasks on various topics within their field of expertise. Objective measures and concomitant verbal protocols were used to assess their behavior and performance. Despite a variable knowledge of the PubMed search tool, neuroscience experts were able to find and select in a limited time adequate references for each task. Sixteen expert biologists of matched professional experience, who were regular users of PubMed but not integrative neuroscience specialists, were asked to perform the same search tasks. Despite their lack of knowledge in neuroscience, as demonstrated by their slower and more frequent reading of the instructions for the tasks, non-expert researchers could find and select adequate references with the same efficiency as neuroscience experts. However, differences were observed between the way experts and non experts proceeded, as exemplified by the larger number of keywords included in the requests to PubMed by non-experts. These data suggest that a high level expertise in a broad scientific field like biology can compensate for an often superficial knowledge of online information search tools, even if the participants are not specialists of the specific domain in which references are searched for. For well-defined reference search tasks in neuroscience, a minimal experience in using PubMed and/or online search tools can compensate for the absence of expertise in neuroscience. However, the mistakes made by high-level researchers using PubMed stress the need for inclusion of formal training to online search tools within higher education programs.

One of the main goals of higher scientific education is to teach students how to perform efficient bibliographic and documentary searches. This study was designed to explore how scientific experts extract information related to their research field, and to assess their level of knowledge of the online tools and bibliographic databases designed for that purpose. More precisely, the field of integrative neuroscience was chosen to evaluate the cognitive strategies used by domain experts to retrieve bibliographic references, and to compare their performance with that of researchers of the same level with no particular expertise in neuroscience.

Biology and medical sciences researchers regularly use the PubMed online search engine (http://www.pubmed.gov), which provides access to the MEDLINE bibliographic database. Despite a lack of formal training to this tool, PubMed has become the gold standard of French neuroscientists for work-related information seeking (Vibert, Rouet, Ros, & Ramond, submitted). Sixteen integrative neuroscience experts of doctorate level or above were asked to perform 5 bibliographic search tasks dealing with various topics within their field of expertise. In addition, 16 expert biologists of matched professional experience, who were regular users of PubMed but not integrative neuroscience specialists, were asked to perform the same search tasks. The average age and number of years of experience using online bibliographic search tools and PubMed for work were similar between both groups of participants. Each participant had 15 minutes to perform each task, and concomitant verbal protocols were used. Participants could take as much time as they wanted to read the instructions for each task, and could read them again at will during the task, but these instructions were hidden from view during the search phases. The tasks were written in both French and English and were precisely defined, as shown by the two following examples: “Find 1 review article dealing with the cellular mechanisms of action of alcohol on the neurons of the central nervous system” (task 1) and “Find 2 articles published in 2004 dealing with the links between neurogenesis and Alzheimer’s disease” (task 4). The order in which the 5 tasks were performed was randomized, and then each participant had to fill in a questionnaire about his knowledge of the functionalities of the PubMed search engine and online database.

The data revealed that as expected, neuroscience experts were able to find and select in 15 minutes adequate references in response to each task. Surprisingly, the biology researchers that regularly use PubMed, but are not experts in neuroscience (non-experts group) were able to find and select adequate references with the same efficiency as neuroscience experts. In addition, the bibliographic search efficiency did not depend on the task, whatever the specificity of the vocabulary or the number of keywords used in its definition. The performance of participants was not related either to their knowledge of the PubMed tool, which was extremely variable between individuals since most of the researchers tested were never formally trained in the use of online bibliographic databases. Among the neuroscience experts, who included PhD students and post-doc researchers as well as junior and senior researchers (age range 25 to 68 years), there was no relationship between the level of professional experience of the participants and their knowledge of the PubMed tool or their bibliographic search performance. The level of knowledge of PubMed functionalities was also not related to the number of years of experience of the participants with the tools.

However, differences were observed between the way experts and non experts proceeded to find and select adequate references for the tasks. Indeed, non-experts in neuroscience took a significantly longer time to read the instructions for the tasks, and asked more often to read them again while they were performing the search tasks. In addition, the non-experts included more keywords in the first request they sent to the PubMed search engine, and opened for some tasks more reference abstracts than the neuroscience experts. This last result and the analysis of the verbal protocols suggest that the non-experts had more difficulties selecting adequate references as responses to the task.

Both neuroscience experts and non-experts made a number of errors using the PubMed interface, but the nature of these mistakes depended on the expertise of the participants. Indeed, the neuroscience experts made significantly more spelling mistakes than the non-experts when they entered keywords to compose their request. Since most of the participants used almost exclusively keywords that were included in the descriptions of the tasks, this result suggests that neuroscience experts do not read those descriptions as precisely as the non-experts.

Altogether, these data suggest that a high level expertise in a broad scientific field like biology can be used to compensate for an often superficial knowledge of online information search tools, even if the participants are not specialists of the specific domain in which references are searched for. Thus, for well-defined reference search tasks in neuroscience such as the ones used in this study, a minimal experience in using PubMed and/or other online search tools can compensate for the absence of expertise in neuroscience. The high number of mistakes that high-level biology researchers make when they use the PubMed tool points the need for inclusion of formal training to online information search tools within higher education programs.