I aim to conduct use-inspired basic research into human cognition and learning. I research how people learn (and think they learn) concepts and categories, with an eye to improving educational and training practices in forensics, medicine, and the natural sciences. At the core of these seemingly diverse fields of study is the process of generalising from one instance to another: a new instance of that finger, person, disease, biological species, statistical concept, or geological structure.

To become an expert with the concepts and categories at hand, novices must learn their basic relational structure—the family resemblance that emerges across many different instantiations. Better understanding how people generalise from their prior experiences to new examples and contexts is fundamental to designing instructional methods that help learners efficiently form new concepts, and correct old ones. The ultimate goal of my research is to develop a domain-general theoretical framework for efficiently creating expertise with concepts and categories.

The Nature of Perceptual Expertise

For my PhD, I researched the nature and development of perceptual expertise in the context of fingerprint identification. Fingerprint examiners spend their days visually comparing pairs of fingerprints, and judging whether they belong to the same finger (e.g., Smith’s right thumb) or two different fingers (e.g., Smith and Jones’s right thumb). This perceptual discrimination task poses a real challenge for novices, and expert examiners with years of experience are not infallible (Tangen, Thompson & McCarthy, 2011). Mismatching fingerprints can often look highly similar due to the use of computer algorithms that help speed up the search for known candidates. Conversely, matching prints can look very different, due to variation in surface, positioning, pressure, movement, and moisture as a print is left behind. 

I compared the performance of fingerprint experts and novices on a series of cognitive tasks, and tracked the performance of fingerprint trainees as they gained experience working in a fingerprint unit. The results of my experiments suggest that a memory for prior instances underlies fingerprint comparison expertise (Searston, Tangen & Eva, 2016). That is, fingerprint experts seem to make use of their vast experience with how prints tend to look and vary to help resolve novel cases. For example, using a novel person discrimination task, we found that fingerprint experts are significantly more accurate than novices at detecting structural or stylistic information across a person’s fingers (Searston & Tangen, in press; blog post). Fingerprint comparison expertise also develops with, and is constrained by, experience in the domain (Searston & Tangen, under review)—as with other forms of perceptual expertise. However, it is surprisingly flexible to changing task demands (Searston & Tangen, 2017). Contrary to the common assumption that image comparison relies almost exclusively on a deliberative perceptual process of aligning features in the case at hand, these findings provide compelling evidence that expert image comparison, in the domain of fingerprints, is marked by an ability to retrieve and flexibly use information distributed across prior instances.

As a postdoc at UQ, I extended my research with fingerprints to other natural categories, including art (e.g., Cubist and Impressionist paintings), ornithology (e.g, hawks, and owls), and faces (e.g., females and males). Together with colleagues, I tested people’s ability to discriminate, remember, and learn these categories with varying amounts of within-image (e.g., resolution, noise), within-category (e.g., Monet paintings), and between-category (e.g., Impressionist paintings) information. One key finding was that people are able to discriminate and remember natural categories with surprisingly little within-image information (Searston, Tangen, & Thompson, in prep).  

Creating Expertise with Concepts and Categories

As a McKenzie Fellow at the University of Melbourne, I am now working to better understand how people learn concepts and categories across several different scientific disciplines (e.g., forensics, geology, biology). In one line of experiments, I’m investigating how testing versus study affects natural category learning (Searston, Tangen, Lodge & Zhen, in prep). Together with colleagues, I’m also examining ways to focus learners on abstract rule-based information and particular instance-based information for efficiently turning novices into fingerprint experts.