The method of educational assessment affects children\u2019s neural processing and performance: behavioural and fMRI Evidence

Steven J Howard,Sahar Bokosmaty,Alysha Calleia,Lisa Kervin,Hana Burianová,Samuel Fynes-Clinton

doi:10.1038/s41539-017-0010-9

Abstract

Standardised educational assessments are now widespread, yet their development has given comparatively more consideration to what to assess than how to optimally assess students’ competencies. Existing evidence from behavioural studies with children and neuroscience studies with adults suggest that the method of assessment may affect neural processing and performance, but current evidence remains limited. To investigate the impact of assessment methods on neural processing and performance in young children, we used functional magnetic resonance imaging to identify and quantify the neural correlates during performance across a range of current approaches to standardised spelling assessment. Results indicated that children’s test performance declined as the cognitive load of assessment method increased. Activation of neural nodes associated with working memory further suggests that this performance decline may be a consequence of a higher cognitive load, rather than the complexity of the content. These findings provide insights into principles of assessment (re)design, to ensure assessment results are an accurate reflection of students’ true levels of competency.

Highlights

National and international programs of standardised educational assessment are widespread
Australia’s National Assessment Program–Literacy and Numeracy (NAPLAN), which is administered annually to students at four points in their schooling, was designed to assess the “sorts of skills that are essential for every child to progress through school and life, such as reading, writing, spelling and numeracy.”[4]. In support of this assertion, there is ample evidence that subsequent academic and life outcomes are strongly predicted by literacy and numeracy domains commonly assessed by these tests.[5,6,7,8]
Imposes differing and extraneous cognitive demands, it was cloze dictation engaged bilateral lingual gyrus, fusiform gyrus, expected that: (a) children’s spelling performance would decrease caudate nucleus, medial frontal gyrus, and anterior cingulate with the increasing cognitive load of assessment; and (b) methods of During the spelling phase, all conditions activated a common testing involving higher cognitive load would recruit brain pattern, which comprised the anterior and posterior areas of the frontoparietal network that are associated with cingulate gyri, bilateral inferior parietal lobule

Summary

INTRODUCTION

National and international programs of standardised educational assessment are widespread. In line with the proposal that the method of assessment active during response inhibition, processing of semantic verbal information, accessing of word meaning during reading, and binding of highly processed perceptual inputs.[23, 24] In contrast, imposes differing and extraneous cognitive demands, it was cloze dictation engaged bilateral lingual gyrus, fusiform gyrus, expected that: (a) children’s spelling performance would decrease caudate nucleus, medial frontal gyrus, and anterior cingulate with the increasing cognitive load of assessment (such that proofreading would impose the highest cognitive load and cortex, reflecting the monitoring of verbal fluency and identification and recognition of words.[25,26,27] multiple choice the lowest cognitive load); and (b) methods of During the spelling phase, all conditions activated a common testing involving higher cognitive load would recruit brain pattern, which comprised the anterior and posterior areas of the frontoparietal network that are associated with cingulate gyri, bilateral inferior parietal lobule (angular and working memory and increased attention (e.g., prefrontal and parietal cortices).[20, 21] To evaluate these hypotheses, behavioural supramarginal gyri), precuneus, insula, parahippocampus, hippocampus, fusiform gyrus, inferior frontal gyrus, and lingual gyrus, analyses and associated neural correlates are reported. Accounting for 70.91% of covariance in the data (see Fig. 3)

DISCUSSION

Participants

Findings

Procedure