Doctoral School of Psychology (Cognitive Science) 

BME TTK, Department of Cognitive Science

Supervisor: Dr. Racsmány Mihály

Behavioral and electrophysiological investigation of retrieval-based learning mechanisms

Introducing the research area

One of the most common misconceptions regarding memory and learning is that the more we re-read the information we would like to memorize, the more it will stick to our memory. In reality, this learning strategy is only effective for the short term, as we might find out when we are barely able to recall the material we learned for a test even after a few days after an exam. To be able to actually memorize the to-be-learned information on the long run, the best method is to quiz ourselves and test our memory of the learned material. This strategy could be summarized by the phenomenon called the testing effect, showing that repeated retrieval facilitates further retrieval compared to restudying. The aim of our study was to better understand the mechanisms behind the testing effect, which not only has theoretical significance, but could also largely benefit educational practice.

 

Brief introduction of the research venue

In the Learning and Memory Laboratory of the Budapest University of Technology and Economics, Department of Cognitive Science we investigate memory processes such as retrieval, memory consolidation and the beneficial effect of testing on long-term memory. The recording of the electrophysiological (EEG) data took place in the Electrophysiology Laboratory of the Department of Cognitive Science.

 

History and context of research

Understanding the functioning and decline of human memory along with its rehabilitation possibilities are becoming a more and more crucial problem both for education and our aging society. One of the greatest findings in learning and memory research was that the role of retrieval is not restricted to assessing previously learned information, but it has an effect on the learning process itself. Repeated retrieval itself is an effective learning strategy that promotes better long-term memory compared to re-reading the material. This facilitating effect of retrieval on long-term memory is called the testing effect [1]. Although the testing effect is a well-known phenomenon in memory research, it is not widely used in educational practice to aid learning [2]. Compared to other learning strategies, testing is highly effective [3], thus gaining a better understanding of its underlying mechanisms bears not only theoretical, but also practical significance. There are many existing theories explaining the relationship between repeated retrieval and long-term memory [4, 5, 6, 7], however, currently none of the theoretical accounts describing the background of the testing effect is widely accepted.

Looking closely at the literature of testing effect, we can find strong similarities between repeated retrieval practice and skill learning. Similarly to skill learning, retrieval-based practice results in lower forgetting rate [1] and makes the memory more resistant to interference effects [8]. Moreover, repeated [9] and spaced [10] practice are proved to be more effective in case of both retrieval practice and skill learning. As a result of automatization processes occurring during repeated practice, skill learning leads to attentional control processes being less involved in retrieval [11]. Retrieval-based practice also results in less attentional involvement in the response processes [12], together with the decreasing activity of the prefrontal cortex [13], a brain area associated with control functions. The parallels described above give rise to the idea that repeated retrieval practice leads to the automatization of retrieval and that this automatization process could be at least one of the factors behind the beneficial effect of testing on long-term memory [14].

 

Aim of the research

An additional characteristic of automatic processes is that in the course of repeated practice, response time is becoming shorter, and this speed-up is characterized by a power-function („power law of practice”; [11]). Since this allows us to describe the degree of automatization with the speed-up of the response, our first experiment [14] was conducted to investigate how the decrease of the reaction time during repeated retrieval practice fits to this power function, and whether the fit of the two is related to long-term retention.

 

Electrophysiological studies of retrieval practice could be very informative of the neurological background of the testing effect, since they allow us to compare the changes distinctive to restudy and retest at the time of the practice phase. However, previous electrophysiological studies of the testing effect used experimental designs with only one practice cycle [e.g.: 15, 16]. As one of the most important facilitating factor of practice is repetition, our second experiment was conducted to explore the changes in event-related potentials (ERP) throughout repeated testing and restudy.

 

Methods

To investigate the effects of retrieval and restudy practice on long-term memory, most recent studies used the following paradigm [2]. The participants first try to memorize the stimuli (in our case Swahili-Hungarian word pairs) which could be presented repeatedly in some of the studies. The initial learning is followed by a practice phase where half of the word pairs are presented again for restudying (“restudy”) and the participants’ memory on the other half is assessed (“test”). In our experiments, we used an associative recall task for the testing condition, where participants are required to recall the Hungarian associates of the presented Swahili words. The repeated practice phase is followed by a retention interval, after which all word pairs are tested in a final test phase, generally using the same task as in the testing condition of the practice phase (Figure 1).

 

 

Figure 1: The experimental procedure of our first experiment. The second experiment only differed in the number of the used stimuli (60 word pairs in total) and in the length of retention interval after the initial learning phase (30 sec).

 

In our first experiment that focused on reaction time during practice, we used 40 word pairs as stimuli. In our second experiment using EEG, participants were asked to learn 60 Swahili-Hungarian word pairs. The initial learning phase of both experiments consisted of 5 consecutive cycles, where each word pair was presented for 5 seconds. The learning phase was followed by a retention interval of 5 minutes in the first, and 30 minutes in the second experiment. The practice phase consisted of 6 restudy and 6 test blocks, where participants had 8 seconds to restudy or retrieve the items. In our first experiment, the reaction time of the responses was recorded as well by asking participants to press the space button when the right answer came to their mind. In the second experiment, participants gave their responses orally to avoid the confounding effect of motoric noise caused by typing. The final test phase took place after a one week retention interval in both experiments.

 

The ERP data of the second experiment was recorded using a 32 channel EEG system, contrasting the evoked responses recorded in the first and second halves of the practice phase in the analysis.

 

The number of participants in the first experiment was 39, whereas in the second experiment 23 healthy Hungarian undergraduates took place, aged between 19 and 29.

 

 

Results

In the final test, subjects showed superior long-term memory performance following a one week retention interval. This replicated the results of previous studies showing that retrieval practice is beneficial for the long-term retention of memories. This benefit of testing was not only apparent in the recall performance, but previously tested information was retrieved faster than restudied information as well (Figure 2).

 

 

Figure 2; based on [14]: The average recall rate (left) and reaction time (right) of restudied and tested word pairs in the final test phase of the first experiment. Error bars represent the standard errors of the mean. Significance levels: ***: p<0.001; **: p<0.01

 

 

We can learn more about the underlying mechanisms of the beneficial effect of retrieval practice by examining the changes throughout the practice phase. The results of our first experiment showed that during retrieval practice, the reaction time of the correct responses  decreased in a pattern that followed a power function [11] similar to skill-learning.

The power function not only showed a close fit to the reaction time data, but the fit (taken as a degree of automatization) was negatively correlated to the recall performance of retrieval-practiced items on the final test. That is, the more closely the power function characteristic of skill learning described the decrease in reaction time, the better the long-term retention of the information became. This relationship did not occur for all learned word pairs when we performed a partial correlation analysis for the overall recall accuracy without the retested items (Figure 3). Based on this, we can say that the positive relationship between the degree of automatization during practice and the retrieval performance on the final test is only present for word pairs practiced with testing [14].

Inspecting these findings one might raise the suspicion that they are simply the result of the automatization of the repeated pressing of the space button. To investigate this prospect, we conducted two additional experiments, which showed that the pattern described above was present even without the button-pressing during the practice phase (in these experiments we compared the reaction times of items practiced for different number of cycles only in the final test). These additional results supported our hypothesis that the observed changes in reaction time is not simply due to motoric speed-up [14].

 

 

Figure 3; based on [14]: The decreasing reaction time of the correct responses of tested items in the practice phase of the first experiment, and the power function fitted to the data (left). The correlation between the extent of fit to the power function (measure of automatization) and the recall performance on the final test (right). Error bars represent the standard errors of the mean. SSE: Sum of Squared Errors.

 

The analysis of our EEG recordings during practice showed changes in three retrieval-related ERP components during retrieval practice.  Compared to the first half of the practice phase, the amplitude of the ERP component related to successful episodic retrieval (late positive component - slow positive wave present for 500‒900 ms after stimulus presentation over left parietal areas; [17, 18]) was more positive going (Figure 4/A). This component is associated with conscious retrieval where information about the study event is also retrieved.

One of the components observed later in the retrieval process and associated with extended, controlled evaluation of retrieval context information (late posterior negativity ‒ a negative wave present from 7‒900 ms until response after stimulus onset over posterior areas [19, 20]) was less negative going in the later section of practice. This points to the conclusion that during repeated practice, the controlled context monitoring processes related to this component become less and less involved. This change was also present during restudy practice (Figure 4/C).

Furthermore, a component that some previous studies associated with general monitoring processes  (late right frontal component - present form 500 ms up until 1400 ms over right frontal areas [21]) had a more negative going amplitude in the second half of retrieval practice (Figure 4/B).

 

 

Figure 4:  The locations of the electrodes used in the second experiment and the amplitude change in the retrieval-related ERP curves during restudy and retrieval practice. A: Late positive component, B: Late right frontal component, C: Late posterior negativity. Significance levels: ***: p<0.001; **: p<0.01; *: p<0.05

 

All in all, our behavioral results support the hypothesis that the beneficial effect of retrieval practice on long-term memory could be mediated by automatization processes. In our EEG results we also observed changes in certain ERP components which points to a conclusion that repeated retrieval is associated with the decreasing involvement of control functions that is one of the key attributes of automatization processes.

 

Possible impacts, further plans

Our experiments could play a significant role in understanding the role of retrieval in learning processes and its underlying mechanisms. Our results support the hypothesis that the beneficial effects of repeated testing are ‒ at least to some extent ‒ the results of automatization processes similar to skill learning. In other words, repeated retrieval of memories results in easier and faster long-term retention, just like going through well-practiced motions. Our behavioral results are just being published [14].  Our EEG experiment (to our best knowledge) is the first electrophysiological study of the testing effect using a paradigm with repeated practice cycles. Our future plans include further electrophysiological experiments based on our current EEG results using paradigms (such as recognition tasks) more suitable for a finer differentiation in retrieval processes.

 

Publications, references, links

 

Publications and presentations

 

●     Racsmány M., Szőllősi Á., & Bencze D. (in press) Retrieval Practice Makes Procedure from Remembering: An Automatization Account of the Testing Effect. Journal of Experimental Psychology: Learning, Memory, and Cognition, http://dx.doi.org/10.1037/xlm0000423

●     Bencze D., Németh K., Szőllősi Á., Racsmány M. (2017) Az epizodikus előhíváshoz köthető elektrofiziológiai komponensek vizsgálata. Megjelent: Személyes tér – Közös világ; A Magyar Pszichológiai Társaság XXVI. Országos Tudományos Nagygyűlése; Kivonatkötet; June 1‒3, 2017., Szimpózium 39 (“Miért a teszt a leghatékonyabb tanulási forma?”), pp. 164–165

●     Szőllősi, Á., Bencze, D., Pajkossy, P., Keresztes, A., & Racsmány, M. (2017). Az előhívás védelmet nyújt a későbbi tanulás zavaró hatásai ellen. Magyar Pszichológiai Társaság XXVI. Országos Tudományos Nagygyűlése (presentation). Szeged, Hungary, June 1‒3, 2017

 

Links

 

http://cogsci.bme.hu - BME Department of Cognitive Science

 

http://www.cogsci.bme.hu/~ktkuser/learningmemory/ - BME KTT - Learning and Memory Research Group

 

http://www.nytimes.com/2011/01/21/science/21memory.html?_r=2&emc=eta1& - New York Times - To Really Learn, Quit Studying and Take a Test

 

References

 

1.    Roediger, H. L., & Karpicke, J. D. (2006a). Test-enhanced learning: taking memory tests improves long-term retention. Psychological Science, 17, 249-255.

2.    Nunes, L.D., & Karpicke, J.D. (2015). Retrieval-based learning: Research at the interface between cognitive science and education. Emerging Trends in the Social and Behavioral Sciences (pp. 1-16). John Wiley & Sons, Inc.

3.    Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students’ learning with effective learning techniques promising directions from cognitive and educational psychology. Psychological Science in the Public Interest, 14(1), 4-58.

4.    Karpicke, J. D., Lehman, M., & Aue, W. R. (2014). Retrieval-based learning: An episodic context account. In B. H. Ross (Ed.) Psychology of learning and motivation, Vol. 61. (pp. 237-284). New York, NY: Elsevier

5.    Carpenter, S. K. (2009). Cue strength as a moderator of the testing effect: The benefits of elaborative retrieval. Journal of Experimental Psychology: Learning, Memory and Cognition, 35, 1563-1569.

6.    Bjork, R. A. (1975). Retrieval as a memory modifier. In R. Solso (Ed.), Information processing and cognition: The Loyola Symposium (pp. 123–144). Hillsdale, NJ: Erlbaum.

7.    Kornell, N., Bjork, R. A., & Garcia, M. A. (2011). Why tests appear to prevent forgetting: A distribution-based bifurcation model. Journal of Memory and Language, 65, 85-97.

8.    Racsmány, M., & Keresztes, A. (2015). Initial retrieval shields against retrieval-induced forgetting. Frontiers in Psychology, 6, 657.

9.    Rawson, K. A., & Dunlosky, J. (2011). Optimizing schedules of retrieval practice for durable and efficient learning: How much is enough? Journal of Experimental Psychology: General, 140(3), 283.

10.  Butler, A. C., Karpicke, J. D., & Roediger III, H. L. (2008). Correcting a metacognitive error: feedback increases retention of low-confidence correct responses. Journal of Experimental Psychology: Learning, Memory, and Cognition, 34(4), 918

11.  Logan, G. D. (1988). Toward an instance theory of automatization. Psychological Review, 95, 492-527.

12.  van den Broek, G. S., Takashima, A., Segers, E., Fernández, G., & Verhoeven, L. (2013). Neural correlates of testing effects in vocabulary learning. NeuroImage, 78, 94-102.

13.  Keresztes, A., Kaiser, D., Kovács, G., & Racsmány, M. (2014). Testing promotes long-term learning via stabilizing activation patterns in a large network of brain areas. Cerebral Cortex, 24, 3025-3035.

14.  Racsmány M., Szőllősi Á., & Bencze D. (accepted for publishing) Retrieval Practice Makes Procedure from Remembering: An Automatization Account of the Testing Effect. Journal of Experimental Psychology: Learning, Memory, and Cognition, http://dx.doi.org/10.1037/xlm0000423

15.  Rosburg, T., Johansson, M., Weigl, M., and Mecklinger, A. (2015). How does testing affect retrieval-related processes? An event-related potential (ERP) study on the short-term effects of repeated retrieval. Cogn. Affect. Behav. Neurosci. 15, 195–210

16.  Bai, C. H., Bridger, E. K., Zimmer, H. D., & Mecklinger, A. (2015). The beneficial effect of testing: an event-related potential study. Frontiers in behavioral neuroscience, 9.

17.  Friedman, D., and Johnson, R. (2000). Event-related potential (ERP) studies of memory encoding and retrieval: a selective review. Microsc. Res. Tech. 51, 6–28

18.  Düzel E, Yonelas AP, Mangun GR, Heinze H-J, Tulving E. (1997). Event-related brain potential correlates of two states of conscious awareness in memory. Proc Natl Acad Sci 94:5973–5978

19.  Johansson, M., & Mecklinger, A. (2003). The late posterior negativity in ERP studies of episodic memory: Action monitoring and retrieval of attribute conjunctions. Biological Psychology, 64, 91–117.

20.  Herron, J. E. (2007). Decomposition of the ERP late posterior negativity: Effects of retrieval and response fluency. Psychophysiology, 44, 233–244.

Hayama, H. R., Johnson, J. D., & Rugg, M. D. (2008). The relationship between the right frontal old/new ERP effect and post-retrieval monitoring: Specific or nonspecific. Neuropsychologia, 46, 1211–1223.