Developing Minds in the Digital Age

Chapter 21. The Australian Science of Learning Research Centre

John Morris

Pankaj Sah

Abstract

The Australian Science of Research Learning Centre (SLRC), established as a special initiative of the Australian Research Council in 2008, brings together neuroscientists, psychologists and educators to work collaboratively towards improving classroom outcomes. Research at the SLRC is organised into three broad themes: “understanding learning”, which focusses on the basic neurophysiological and psychological processes involved in learning; “measuring learning”, which aims to develop novel techniques to quantify learning as it occurs dynamically in the classroom; and “promoting learning”, which takes the insights of the other themes to formulate novel strategies to improve learning outcomes. SLRC projects have examined such phenomena as mathematics anxiety, the role of feedback and Bayesian approaches to learning and the brain. The chapter also addresses recent scepticism about the utility of using neuroscientific insights to improve classroom learning.

Introduction

The Australian Science of Learning Research Centre (SLRC) was established in 2013 with a mission to investigate how advances in our basic understanding of the neural and psychological processes underlying learning and memory can help improve educational outcomes in the classroom. This is an attractive project from a governmental perspective: it co-opts high profile, state of the art neuroscientific research (e.g. brain imaging, etc.) in the service of addressing a bread and butter concern for every family, i.e. schooling. Several voices, however, have expressed scepticism about whether this political enthusiasm for educational neuroscience will be matched by any tangible benefits, (Bowers, 2016[1]; Bruer, 1997[2]; Cubelli, 2009[3]). This chapter will examine the historical context and the specific aims and achievements of the SLRC. It will also address the sceptical challenges that have been raised to the application of neuroscientific findings to classroom teaching, and attempt to outline how the collaboration between neuroscience, psychology and education can be both productive and mutually beneficial.

History of educational research

While academic interest in teaching methods date back at least as far as the Socratic elenchus (Plato, 1980[4]), modern scientific approaches to pedagogy date from the establishment of education departments in American universities at the end of the 19th century. The pioneers of psychological research into learning – Ebbinghaus (1885[5]), Thorndike (1976[6]), Pavlov (1927[7]) and Skinner (1938[8]) – had a profound effect on scientific approaches to education. However, despite the adoption of these scientific methods from psychology, educational research has been widely criticised for significant conceptual and methodological problems (Higgins and Simpson, 2011[9]; Kaestle, 1993[10]; Lagemann, 1997[11]; Vinovskis, Kaestle and Glennan, 2000[12]). Lagemann (1997[11]), in a wide-ranging review, suggested that “new, more collegial patterns of cooperation” should be encouraged between “scholars of education, scholars in other fields and disciplines, school administrators and teachers” to foster “knowledge-based reform in education”. It was in this historical context that initiatives to bring neuroscientific insights to education were first instituted.

The international “Brain and Learning” project at the OECD’s Centre for Educational Research (OECD, 2002[13]) involving various international institutions, including the Sackler Institute (United States), University of Granada (Spain), RIKEN Brain Science Institute (Japan) National Science Foundation (United States), the Lifelong Learning Institute (United Kingdom) and INSERM (France), brought together neuroscientists, psychologists and educationalists to foster collaboration in improving teaching and learning. The success of the OECD Brain and Learning project led to the creation of dedicated multi-institutional centres for educational neuroscience in Cambridge (2005) and London (2008). The Australian Science of Learning Centre (SLRC) was established in 2013 as a special initiative of the Australian Research Council.

Australian SLRC

Administered by the University of Queensland, and in partnership with three state education departments and four international universities, the SLRC involves a collaboration between the University of Melbourne, the University of New England, Macquarie University, Flinders University, Deakin University, Curtin University and the Australian Council for Educational Research (ACER). Facilities available at the SLRC include the Educational Neuroscience Classroom at the University of Queensland, which allows electroencephalography (EEG), eye tracking and other psychophysiological measurements on multiple subjects during learning sessions, and the Learning Interaction Classroom at the University of Melbourne which can accommodate 30 students in a conventionally structured classroom equipped with multiple channels of high definition video capture and the ability to collect psychophysiological data from wireless devices. Research at the SLRC is organised into three broad themes: understanding learning, measuring learning and promoting learning. The understanding learning theme focuses on exploring fundamental neurophysiological and psychological processes underlying learning and memory; the measuring learning theme focusses on developing novel techniques to measure learning dynamically, as it is happening, in classrooms and digital learning environments; finally, the promoting learning theme aims to develop novel strategies, on the basis of insights gained in the other themes, to improve learning outcomes.

Minding the gap

Despite the founding of organisations, such as the SLRC, and increasing political and academic interest in educational neuroscience, there has also been notable criticism of this new discipline. Bruer (1997[2]) argues that the conceptual gap between neuroscience and education is “a bridge too far”, and that neuroscientific findings cannot make a meaningful contribution to classroom practice. Cubelli (2009[3]) agrees that the aim of educational neuroscience is misguided and proposes that advances in psychology rather than neuroscience are potentially useful to classroom teaching. Nineteen years after Bruer’s article (1997[2]), Bowers (2016[1]) reinforces and amplifies its sceptical argument, concluding that that “there is no reason to assume that neuroscience will add any new insights relevant to teaching” and that neuroscience “is (and always likely will be) irrelevant to the task of designing or evaluating instruction”. These criticisms, which clearly question the entire raison d’être of the SLRC and educational neuroscience in general, are addressed in detail in the remainder of this chapter, using examples of research from the SLRC and other groups.

One project in the SLRC’s measuring learning theme is investigating mathematics anxiety and its effect on high school students (Buckley et al., 2016[14]). Students with mathematics anxiety have previously been shown to have greater activation of the amygdala, a region of the brain strongly implicated in emotional processing, particularly fear learning and memory (Young, Wu and Menon, 2012[15]). The neural circuitry underlying fear conditioning (aversive Pavlovian conditioning) is also a principal focus of research in the SLRC’s understanding learning theme. A potential role for the amygdala in mathematics anxiety might be considered a good example, therefore, of how neuroscientific knowledge can contribute to teaching practice. Bowers (2016[1]), however, rejects claims that neuroscientific insights into the role of the amygdala are of any benefit to educational practice: he states that “everyone knows that stressed and fearful students make poor learners”, and that therefore information concerning the underlying neural circuitry is “trivial” and can contribute nothing to classroom teaching. This conclusion, however, ignores the fact that improving our understanding of amygdala function may also transform the way we conceptualise interactions between emotion and learning at a psychological level. This new conceptual framework might then facilitate the development of novel techniques to mitigate mathematical anxiety that might not otherwise have become apparent without the neuroscientific input.

Another SLRC project is investigating the partial reinforcement extinction effect (PREE), a paradoxical phenomenon in which the unexpected omission of reinforcement during conditioning leads to persistence of conditioned responding in extinction (Amsel, 1962[16]; Capaldi, 1966[17]). The PREE can therefore be seen as a model for enhanced memory retention. The SLRC PREE project is using immunohistochemistry, in vivo electrophysiology in animal subjects and electroencephalography (EEG) in human subjects during partially reinforced fear conditioning, to identify and characterise the neural responses and brain circuitry mediating the PREE, particularly in the hippocampus and amygdala (Morris, 2015[18]). Although the PREE is difficult to reconcile with standard associative learning models (Rescorla and Wagner, 1972[19]), it is predicted by Bayesian models, which emphasise the role of uncertainty and surprise in learning (Courville, Daw and Touretzky, 2006[20]). The application of Bayesian models of brain function in an educational context has been advocated by several authors (Fischer, 2009[21]). Critics, however, have explicitly rejected the idea that a Bayesian approach to the role of surprise and uncertainty in learning can be usefully applied to classroom teaching. Bowers (2016[1]) states that “the link between Bayesian models of cortical computation and education is hard to see”. However, several current classroom projects at the SLRC do indeed lend themselves to a Bayesian interpretation, indicating, therefore, that such criticism is narrow and short-sighted.

A third project is investigating how “confusion” can paradoxically enhance learning by evoking greater engagement with the material to be learned (D’Mello and Graesser, 2014[22]; Pachman et al., 2016[23]). Behavioural responses (e.g. facial expression, facial EMG, eye fixation, posture, learner-computer interactions, etc.) and physiological responses (e.g. skin conductance, heart rate, pupillometry, brain imaging, etc.) are being investigated as potential measures of confusion to help the development of predictive models for real world learning situations (Pachman et al., 2016[23]). Theoretical accounts of confusion relate it to cognitive disequilibrium or dissonance, a psychological theory that proposes that individuals try to minimise conflicts that arise between new information and their existing beliefs (Festinger, 1957[24]). In this view, as a result of the conflict, incoming information undergoes enhanced processing, and the new information is either rejected or incorporated into a revised set of beliefs (i.e. learned). This effect of “confusion” also lends itself very readily to a Bayesian interpretation: according to the Bayesian brain hypothesis, the brain is essentially a statistical prediction machine, continually predicting stimulus events according to a “world model” or “prior beliefs” (Courville, Daw and Touretzky, 2006[20]). Unexpected or surprising events lead to an update of these prior beliefs according to Bayes’ rule (Courville, Daw and Touretzky, 2006[20]). Bayesian models could provide, therefore, a unified account of uncertainty and surprise at the neural and psychological level, and of confusion at the classroom teaching level. This account can then drive changes in the response to when students are confused. The key is to be aware of when confusion is apparent and respond accordingly.

In the classroom, it has been apparent for many years that a key determinant of effective learning is feedback (Hattie and Donoghue, 2016[25]). The role of feedback in computer-based, intelligent learning environments (ILEs) is also being investigated at the SLRC, (Holland and Schiffino, 2016[26]; Timms, DeVelle and Lay, 2016[27]). Physiological measures such as heart rate and skin conductance, as well as eye tracking, pupillometry and EEG are being used to dynamically assess and optimise the role of feedback in ILEs. The role of feedback in learning is informed by the psychological and neuroscientific concept of prediction error (Timms, DeVelle and Lay, 2016[27]). Influential models of learning propose that the prediction error “mismatch” between expected and actual outcomes is the primary driver of learning (Rescorla and Wagner, 1972[19]). This role of feedback in learning can again be very easily accommodated by a Bayesian account of the updating of prior beliefs (Courville, Daw and Touretzky, 2006[20]). The conceptual convergence between confusion, cognitive dissonance, feedback, prediction error and the PREE presents an opportunity for the development of novel teaching methods (and teacher feedback), and models to advance our understanding of these learning processes, e.g. using partial reinforcement to elicit or enhance confusion or using feedback to elicit surprise. The multidisciplinary nature of the SLRC allows such potentially convergent models to be studied from the single neuron level (e.g. animal electrophysiology) to the classroom.

Another phenomenon with the prospect of a productive collaboration between education and neuroscience is the testing effect, which refers to the improvement in memory retention that arises from substituting free recall tests for study time during learning (Roediger and Karpicke, 2006[28]). The testing effect has parallels with the neurophysiological phenomenon of reconsolidation, in which retention of conditioned responding can be enhanced or degraded by the unpredictable presentation of unreinforced conditioned stimuli which are hypothesised to return the memory trace to a labile state (Lee, 2008[29]; Pedreira, Pérez-Cuesta and Maldonado, 2004[30]). Critics, however, have rejected claims of parallels (Carew and Magsamen, 2010[31]) between reconsolidation and “the testing effect”, arguing that the neurophysiological processes of consolidation and reconsolidation can make no contribution to understanding the effect or its implementation in the classroom (Bowers, 2016[1]). This criticism ignores the fact that reconsolidation is an active area of current research, not only concerning the underlying neural mechanisms, but also the specific protocols and boundary conditions with which it is expressed. For example, labilisation-reconsolidation has been shown to strengthen declarative memory in humans, but only when two or more reactivations are used, and only when cues alone and not cues and responses are used for reactivation (Forcato, Rodríguez and Pedreira, 2011[32]). The possibility for such neuroscientific findings to directly inform classroom practice is clear from this example. It is also quite possible that novel behavioural findings from classroom studies may guide hypotheses and research at the molecular and cellular level. This potentially productive “two-way street” belies Bowers (2016[1]) contention that neuroscientific findings can never make a positive contribution to education.

Another research area receiving increasing interest from educational neuroscience is the parallel between experimental protocols that induce long-term potentiation (LTP) and the spacing effect (Kornmeier and Sosic-Vasic, 2012[33]). LTP is widely regarded as the most promising cellular model of long-term memory storage (Bliss and Lomo, 1973[34]). Experimental studies across a range of species has identified the critical importance of the distributed timing of stimuli in the optimisation of LTP (Scharf et al., 2002[35]). The psychological “spacing effect”, which describes the positive effect of distributed practice often over days or weeks on memory retention (Ebbinghaus, 1885[5]), has therefore an obvious parallel with optimal LTP timing (Kornmeier and Sosic-Vasic, 2012[33]). Taking into account the optimal timing parameters required for LTP, Kelley and Whatson (Kelly and Whatson, 2013[36]) conducted an innovative classroom study with high school science students in which traditional lesson structures were replaced with “Spaced Learning”. Intensive 20-minute learning periods were separated by 10-minute periods of distractor physical activities, such as juggling or clay modelling. Learning content was repeated after each 10-minute “stimulus-free” gap in order to match the optimal conditions for establishing LTP (Scharf et al., 2002[35]). Kelley and Whatson (2013[36]) report significantly increased learning rates with the LTP-inspired teaching programme, and significantly higher test scores with Spaced Learning compared to traditional methods in a review of course content. The development of the Spaced Learning programme took place over seven years, and crucially involved neuroscientists and psychologists working directly with classroom teachers.

The Spaced Learning project (Kelly and Whatson, 2013[36]) demonstrates how neuroscientific insights into cellular and molecular mechanisms of learning can drive developments in education practice and make a direct and positive impact on classroom teaching. Conversely, insights gained in the classroom application of spaced learning may help to inform future neurophysiological studies of learning, completing a rich and productive “two-way street” of collaboration. For example, significant effects in the classroom relating to the use of multiple stimulus modalities, motivational and emotional manipulations, diurnal and sleep patterns, individual (genetically based) differences in response, etc., could also be studied in animal models. Insights from these studies could then in turn, inform classroom practice, perhaps by suggesting novel approaches or optimal parameters. The multidisciplinary and collaborative structure of organisations such as the SLRC can facilitate this dynamic “two-way street” of communication by bringing together neuroscientists, psychologists and educationalists on common projects.

Critics of educational neuroscience typically rely on creating a sharp divide between psychology and neuroscience. Bowers (2016[1]) argues that psychology has made and will continue to make important contributions to education, principally because it deals in behavioural outcomes, but that neuroscience cannot make a similar contribution since it only deals with neural activity: “changes in brain states are irrelevant for evaluating the efficacy of an instruction”. However, such an argument represents a very blinkered and inflexible approach to levels of understanding in science. Bowers (2016[1]) argues that psychological constructs, such as “episodic memory”, “semantic memory”, “attention”, “phonological processing”, etc., are valid and useful for improving educational practice. Neuroscience not only attempts to describe the neural circuitry and processes underlying these psychological concepts, but by relating them to neurophysiological phenomena such as LTP, consolidation, reconsolidation, oscillatory coherence, etc., offers the real prospect of radically transforming the psychological concepts themselves. Just as advances in biomedical science have transformed ancient terms such as “humours”, “jaundice” and “consumption” allowing more precise medical diagnoses and treatments, so neuroscientifically transformed psychological concepts may facilitate improvements in teaching and learning outcomes.

Conclusion

This chapter has examined in detail a convergence of ideas and projects between single neurons, animal models and the classroom: fear conditioning and mathematics anxiety; uncertainty-dependent PREE and confusion; reconsolidation and the testing effect; and LTP and spaced learning. The Australian SLRC, by bringing together neuroscientists, psychologists and educationalists in collaborative projects, appears well placed to facilitate and exploit these convergences, which offer the prospect not only of improving classroom teaching through novel procedures and materials, but also of providing behavioural data to guide and inform research at the cellular and molecular level. Indeed, learning principals emerging from the SLRC are now being implemented in classrooms in Australia, and longitudinal evaluation of these trials will provide validation. The claim of several critics that this promise is illusory and that neuroscience can never hope to inform educational practice often appears as a narrow and blinkered exercise in circular reasoning, i.e. neuroscientific data is strictly relevant only to neural activity, and therefore can never be applied to classroom teaching.

An analogy between clinical genetics and educational neuroscience can help to refute the claim that basic neuroscience can never be of relevance for classroom teaching. The discovery of the structure of DNA, or even the determining of the human genome, did not immediately change medical practice. The identification of a particular sequence of DNA base pairs, in isolation, cannot improve the clinical care of a patients. However, determining that a particular DNA sequence codes for a protein that alters the susceptibility for a certain disease can dramatically alter classification, diagnosis, prognosis and treatment (Kelly and Whatson, 2013[36]). Importantly, such advances occur when information about a DNA sequence is directly related to a clinical outcome, i.e. in a cross-disciplinary project. There is every reason to expect, therefore, pace Bruer (Bruer, 1997[2]); Cubelli (Cubelli, 2009[3]) and Bowers (Bowers, 2016[1]), that just as molecular genetics is now revolutionising medical practice, so advances in our understanding of the neural basis of learning and memory will also be able to transform educational practice.

References

[16] Amsel, A. (1962), “Frustrative nonreward in partial reinforcement and discrimination learning: Some recent history and a theoretical extension”, Psychological Review, Vol. 69, pp. 306-28, http://www.ncbi.nlm.nih.gov/pubmed/13861049.

[34] Bliss, T. and T. Lomo (1973), “Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path”, The Journal of Physiology, Vol. 232/2, pp. 331-356, http://www.ncbi.nlm.nih.gov/pubmed/4727084.

[1] Bowers, J. (2016), “The practical and principled problems with educational neuroscience”, Psychological Review, Vol. 123/5, pp. 600-612, http://dx.doi.org/10.1037/rev0000025.

[2] Bruer, J. (1997), “Education and the brain: A bridge too far”, Educational Researcher, Vol. 26, pp. 4-16, https://www.jsmf.org/about/j/education_and_brain.pdf (accessed on 19 November 2018).

[14] Buckley, S. et al. (2016), “Understanding and addressing mathematics anxiety using perspectives from education, psychology and neuroscience”, Australian Journal of Education, Vol. 60/2, pp. 157-170, http://dx.doi.org/10.1177/0004944116653000.

[17] Capaldi, E. (1966), “Partial reinforcement: A hypothesis of sequential effects”, Psychological Review, Vol. 73/5, pp. 459-477, http://www.ncbi.nlm.nih.gov/pubmed/5341660.

[31] Carew, T. and S. Magsamen (2010), “Neuroscience and education: An ideal partnership for producing evidence-based solutions to guide 21st century learning”, Neuron, Vol. 67/5, pp. 685-688, http://dx.doi.org/10.1016/j.neuron.2010.08.028.

[20] Courville, A., N. Daw and D. Touretzky (2006), “Bayesian theories of conditioning in a changing world”, Trends in Cognitive Sciences, Vol. 10/7, pp. 294-300, http://dx.doi.org/10.1016/j.tics.2006.05.004.

[3] Cubelli, R. (2009), “Discussion forum theories on mind, not on brain, are relevant for education”, CORTEX, Vol. 45, pp. 562-564, http://dx.doi.org/10.1016/j.cortex.2008.06.006.

[4] Destree, P. and Z. Giannopoulou (eds.) (1980), The Symposium, Cambridge University Press, Cambridge, http://dx.doi.org/10.1017/9781316273166.

[22] D’Mello, S. and A. Graesser (2014), “Confusion and its dynamics during device comprehension with breakdown scenarios”, Acta Psychologica, Vol. 151, pp. 106-116, http://dx.doi.org/10.1016/J.ACTPSY.2014.06.005.

[5] Ebbinghaus, H. (1885), Memory: A Contribution to Experimental Psychology, Dover, New York, NY, http://dx.doi.org/10.5214/ans.0972.7531.200408.

[8] Elliott, R. and B. Skinner (1938), The Behavior of Organisms: An Experimental Analysis, Appleton-Century-Crofts Inc., New York, NY, http://s-f-walker.org.uk/pubsebooks/pdfs/The%20Behavior%20of%20Organisms%20-%20BF%20Skinner.pdf.

[24] Festinger, L. (1957), A Theory of Cognitive Dissonance, Standford University Press, Stanford, CA.

[21] Fischer, K. (2009), “Mind, brain, and education: Building a scientific groundwork for learning and teaching”, Mind, Brain, and Education, Vol. 3/1, pp. 3-16, http://dx.doi.org/10.1111/j.1751-228X.2008.01048.x.

[32] Forcato, C., M. Rodríguez and M. Pedreira (2011), “Repeated labilization-reconsolidation processes strengthen declarative memory in humans”, PloS One, Vol. 6/8, p. e23305, http://dx.doi.org/10.1371/journal.pone.0023305.

[25] Hattie, J. and G. Donoghue (2016), “Learning strategies: A synthesis and conceptual model”, npj Science of Learning, Vol. 1/1, p. 16013, http://dx.doi.org/10.1038/npjscilearn.2016.13.

[9] Higgins, S. and A. Simpson (2011), “Visible learning: A synthesis of over 800 meta-analyses relating to achievement”, British Journal of Educational Studies, Vol. 59/2, pp. 197-201.

[26] Holland, P. and F. Schiffino (2016), “Mini-review: Prediction errors, attention and associative learning”, Neurobiology of Learning and Memory, Vol. 131, pp. 207-215, http://dx.doi.org/10.1016/j.nlm.2016.02.014.

[10] Kaestle, C. (1993), “The awful reputation of education research”, Educational Researcher, Vol. 22/1, p. 23, http://dx.doi.org/10.2307/1177303.

[36] Kelly, P. and T. Whatson (2013), “Making long-term memories in minutes: A Spaced Learning pattern from memory research in education”, Frontiers in Human Neuroscience, Vol. 7, p. 589.

[33] Kornmeier, J. and Z. Sosic-Vasic (2012), “Parallels between spacing effects during behavioral and cellular learning”, Frontiers in Human Neuroscience, Vol. 6, p. 203, http://dx.doi.org/10.3389/fnhum.2012.00203.

[11] Lagemann, E. (1997), “Contested terrain: A history of education research in the United States, 1890–1990”, Educational Researcher, Vol. 26/9, pp. 5-17, http://dx.doi.org/10.3102/0013189X026009005.

[29] Lee, J. (2008), “Memory reconsolidation mediates the strengthening of memories by additional learning”, Nature Neuroscience, Vol. 11, pp. 1264-1266, http://dx.doi.org/10.1038/nn.2205.

[18] Morris, J. (2015), The Neural Basis of the Partial Reinforcement Extinction Effect, PhD thesis, The University of Queensland, Brisbane, http://dx.doi.org/10.14264/uql.2015.390.

[13] OECD (2002), Understanding the Brain: Towards a New Learning Science, OECD Publishing, Paris, https://www.oecd-ilibrary.org/education/understanding-the-brain_9789264174986-en.

[23] Pachman, M. et al. (2016), “Eye tracking and early detection of confusion in digital learning environments: Proof of concept”, Australasian Journal of Educational Technology, Vol. 32/6, http://dx.doi.org/10.14742/ajet.3060.

[7] Pavlov, P. (1927), Conditioned Reflexes: An Investigation of the Physiological Activity of the Cerebral Cortex, Oxford University Press, Oxford.

[30] Pedreira, M., L. Pérez-Cuesta and H. Maldonado (2004), “Mismatch between what is expected and what actually occurs triggers memory reconsolidation or extinction”, Learning & Memory, Vol. 11/5, pp. 579-585, http://dx.doi.org/10.1101/lm.76904.

[19] Rescorla, R. and A. Wagner (1972), “A theory of pavlovian conditioning: Variations in the effectiveness of reinforcement and nonreinforcement”, Classical Conditioning II: Current Research and Theory, Vol. 2, pp. 64-99, https://pdfs.semanticscholar.org/afaf/65883ff75cc19926f61f181a687927789ad1.pdf.

[28] Roediger, H. and J. Karpicke (2006), “Test-enhanced learning”, Psychological Science, Vol. 17/3, pp. 249-255, http://dx.doi.org/10.1111/j.1467-9280.2006.01693.x.

[35] Scharf, M. et al. (2002), “Protein synthesis is required for the enhancement of long-term potentiation and long-term memory by spaced training”, Journal of Neurophysiology, Vol. 87/6, pp. 2770-2777, http://dx.doi.org/10.1152/jn.2002.87.6.2770.

[6] Thorndike, E. (1976), “The contributions of psychology to education”, The Journal of Educational Psychology, Vol. 1, pp. 5-12.

[27] Timms, M., S. DeVelle and D. Lay (2016), “Towards a model of how learners process feedback: A deeper look at learning”, Australian Journal of Education, Vol. 60/2, pp. 128-145, http://dx.doi.org/10.1177/0004944116652912.

[12] Vinovskis, M., C. Kaestle and T. Glennan (2000), The Federal Role in Educational Research and Development, Brookings Institution Press, http://dx.doi.org/10.2307/20067226.

[15] Young, C., S. Wu and V. Menon (2012), “The neurodevelopmental basis of math anxiety”, Psychological Science, Vol. 23/5, pp. 492-501, http://dx.doi.org/10.1177/0956797611429134.