STROOP EXPERIMENT

Stroop Experiment

Abstract

Stroop interference is often taken as evidence for reading automaticity even though young and poor readers, who presumably lack reading automaticity, present strong interference. Here the relationship between reading skills and Stroop interference was studied in a lab room in Queens College. Greater interference was observed in classmates. Moreover, poorer reading skills were found to correlate with greater Stroop interference in the general school population. In correlation and regression analyses, interference was primarily associated with reading speed, with an additional unique contribution of reading accuracy. Color naming errors were few and not comparably related to reading skills. The relation of reading skill to Stroop interference was examined in computational modeling simulations.

Introduction

Skillful reading is considered largely automatic at the word level; the development of automaticity is generally taken to be a major goal of reading instruction and practice. The color-word Stroop interference task is a classic naming task in which color naming is slowed down by an interfering (incongruent) printed word. This interference can be taken as a measure of reading automaticity. Therefore, intuitively, reading skill should be expected to be positively correlated with Stroop interference, with better readers exhibiting stronger interference. Even though expression of such a general expectation is found in the literature, certain research findings also exist that are difficult to reconcile with it. In the present study we test this hypothesis that Stroop interference is positively related to reading skill and discuss the implications both for the notion of automaticity in reading and for the theoretical interpretation of the Stroop task. We present empirical evidence replicating and extending previous reports of a negative relation between Stroop interference and reading skill. We then conduct simulations of the observed findings in two prominent but quite different computational models of the Stroop task, one due to Roelofs, 2003, Roelofs, 2005 and Roelofs and Hagoort, 2002 and the other due to Cohen and colleagues. We use these simulations to address the puzzle of why the relation between Stroop interference and reading ability is negative rather than positive. (Ehri and Wilce, 1979)

Automaticity in reading

Automaticity is a complex notion, generally considered to be a graded feature of task performance related to (a) speed, (b) voluntariness (in initiation, control, and termination), (c) cognitive resource requirements (attention as effort) and (d) conscious awareness (attention as focused concentration). It can be modeled as a process feature or as a transition from algorithmic processing to memory retrieval. Automatic processes are performed rapidly, without conscious intent or guidance, and with little effort, thus allowing the simultaneous performance of other tasks at little or no cost. (Everatt et al., 1999) Skilled reading is considered to be a largely automatic process, at least at the word level, including word decoding, up to lexical access, although text-level automaticity has also been proposed. This means that little effort or attention is required to derive word meanings from the written letter strings. Thus, in skilled readers, sufficient cognitive resources are thought to be available for computation of the meaning of ...