Send us your feedback

Thank you for your feedback. An email has been sent to the ESRC support team.

An error occured whilst sending your feedback. Please review the problems below.

How do readers code letter position?

Grant reference: RES-000-22-3354

« View grant details

Impact Report details

How do readers code letter position?
To cite this output: Davis, C, (2012) How do Readers Code Letter Position? ESRC Impact Report, RES-000-22-3354. Swindon: ESRC

Primary contributor

Author Colin Davis

Additional contributors

Co-author Stephen Lupker


This research has provided further insight into the way in which letter and morpheme position is coded in the brain. 1. We further developed a new methodology, called sandwich priming, that offers a more sensitive test of theoretical models than the standard masked priming methodology. 2. We used this methodology to provide a critical test of current theoretical models of orthographic input coding. 3. We also used this methodology to explore the relative position priming constraint. 4. We used masked priming to explore whether the basic units of representation underlying the initial stages of the visual word identification process are letters or graphemes. 5. The results from these investigations together with many findings from other labs have been synthesised in a major theoretical paper in Psychological Review describing the spatial coding model of visual word identification. 6. We developed user-friendly Windows software to simulate this model, and made it freely available to other researchers. 7. Our investigation of response congruency effects in masked priming provided evidence for selective lateral inhibitory connections. 8. To further test the model, we established a 20-centre collaborative network for masked form priming experiments. Data collection is now underway on the first ever “megastudy” of masked priming. 9. We sought to generalise our findings on letter position coding to the issue of how morpheme position is coded. 10. Exploration of an apparent outlier in the predictions of the spatial coding model led to the discovery of contextual effects in masked priming.

1. Our further demonstrations of the sandwich priming methodology have been described in one published paper (Stinchcombe, Lupker, & Davis, 2012) and one paper currently under review (Davis, Lupker, & McCormick, submitted). 2. Our use of this methodology to adjudicate between different models is described in Davis, Lupker, & McCormick (submitted). 3. Our use of this methodology to explore the relative position priming constraint is described in Stinchcombe, Lupker, and Davis (2012). 4. Our investigation of the basic units of representation underlying the initial stages of the visual word identification process (letters or graphemes) is described in Lupker, Acha, Davis, and Perea (2012). 5. Our computational modelling work and the theoretical model we have developed is described in Davis (2010). 6. The software to simulate this model is available from this site: 7. Data collection from our 20-centre masked form priming megastudy is ongoing. 8. Our work on the response congruency effect and its implications for lateral inhibition is described in a published paper (Loth & Davis, 2011). 9. Our investigation of position coding in morphological processing has been described in two papers (Crepaldi, Rastle, & Davis, 2010; Crepaldi et al., under revision). 10. Our discovery of contextual effects in masked priming has been described at the UKOG meeting in 2010, and is the subject of a paper that is currently in preparation.

We have published peer-reviewed articles in Psychological Review (Davis, 2010), Journal of Experimental Psychology: Human Perception & Performance ( Lupker et al., 2012), Language and Cognitive Processes (Stinchcombe et al., 2012), and Memory and Cognition (Crepaldi et al., 2010). Other papers are in the review process at Journal of Experimental Psychology: Human Perception & Performance (Crepaldi et al., under revision) and Journal of Experimental Psychology: Learning, Memory & Cognition (Davis et al., submitted). Additionally, two chapters have appeared in edited books: (Davis, 2011, edited by J. Adelman; Loth & Davis, 2011, edited by E. Davaelaar). Conference presentations have been given at international conferences including Psychonomics (in Boston, St. Louis, and Seattle), the European Society for Cognitive Psychology (San Sebastian), and the International Morphological Processing Conference (in Turku and San Sebastian) as well as national meetings including the Experimental Psychology Society meeting (London), the UK Orthography Group meeting (in Nottingham and Reading), and the Canadian Society for Brain, Behavior, and Cognitive Science (Nova Scotia). In addition, external talks have been given at several UK Psychology departments, including Birkbeck, Bristol, Imperial, Nottingham, Reading, Warwick, and York.

Our research has had an impact on other researchers in the field of reading research. The attempt to “crack” the neural code underlying reading has been the subject of intense research activity in the last few years. The key methodological technique that has been used by researchers over the last decade has been the masked form priming paradigm. Our development of the sandwich priming technique provides a more sensitive variant of this methodology that is being embraced by researchers in the field and promises to accelerate the progress of research. As one example, Professor Grainger’s group in Marseille have recently employed this methodology in conjunction with ERP recordings, and have shown that sandwich priming greatly amplifies the magnitude of effects that are observed in this paradigm. Thus, our research is also having an impact on researchers in the field of neuroimaging. The publication of the theoretical paper on the spatial coding model (Davis, 2010) has had a clear impact on this field of research. This impact, is illustrated, for example, by the fact that this paper has already been cited over 50 times (according to Google scholar citations), a very high rate of citation for such a recent paper in psychology.

The international network of researchers that we have formed will lead to the first masked priming mega-study, which will provide a stringent test of current models and facilitate the development of improved models. We hope that this initiative will result in further collaborative efforts in which researchers from different labs join together to produce large-scale data sets aimed at addressing significant theoretical questions.

Impacts 8 and 10 were not foreseen at the outset of the project. The starting point in both cases was an attempt to address existing limitations. The development of a computational model that makes quantitative predictions was facilitated by existing datasets, but also revealed the limitations of these datasets (e.g., relatively small sample sizes, variability across labs). Generating a much larger dataset will make it possible to reliably compare many different stimulus conditions, and thereby provide a better test of this and other models. In the second case, the success of the model in explaining a very large number of data points highlighted a particular condition that was not well captured by the model. Further exploration revealed that the model could accurately predict the observed priming effect if its vocabulary was constrained to contain words of the same length as the target word. This insight raised the possibility that participants in the experiment were somehow adapting to the fixed length of the target stimuli. In support of this possibility, we have now conducted several experiments showing that the priming effects for different-length primes are greatly reduced when the length of the target stimuli is not fixed (as it has been in prior experiments). That is, it appears that participants exploit contextual factors to constrain the set of lexical candidates that are considered during the identification process. These examples illustrate how a well-specified computational model can identify critical findings, suggest novel mechanisms, and provide the impetus for new experiments.

This research grant was planned as “basic” research and the expectation was that its main impact would be on other researchers in the field. Our long term objective is to use the knowledge about the underlying mechanisms of the reading process which is gained through this research to help us better understand the nature of reading impairments, particularly those arising from brain damage and/or abnormal reading development, and how those impairments can be remedied. A second objective is to use this knowledge to inform reading education programs, allowing for improved instruction techniques.

Cite this outcome


Davis, Colin and Lupker, Stephen. How do readers code letter position?: ESRC Impact Report, RES-000-22-3354. Swindon: ESRC


Davis Colin and Lupker Stephen. How do readers code letter position?: ESRC Impact Report, RES-000-22-3354. Swindon: ESRC.