BMe Research Grant
BMe Research Grant - 2017
Preventing accidental retrieval of unwanted memories requires active suppression of the undesired content, however, current practice of memory research treats memory inhibition as unitary process. Think/No think task is widely known means to investigate this phenomenon, in which forgetting occurs as a result of the automatic inhibition of retrieval according to our approach. With our modified paradigm we investigated the effect of both successful and unsuccessful stopping an already started retrieval process on the later recall performance. Our results can take us closer to build a nuanced framework of forgetting and may give a deeper understanding on the effects of stressful events and traumatic episodes on memory.
The research took place in the Learning and Memory Research Group of the Budapest University of Technology and Economics, Department of Cognitive Science, with the support of Hungarian Brain Research Program (KTIA NAP 13-2-2014-0020). Our lab is widely involved in investigating retrieval and memory consolidation, the beneficial effect of testing on memory, inhibitory control of memory retrieval as well as memory and executive functions in psychiatric and neurological disorders using the widest range of experimental psychological tools (behavioral paradigms, eye tracking, EEG, fMRI and neuropsychological tests).
Investigation of forgetting and suppression of memories is one the most important pillar to understand the background of damaged memory processes during several neurological and neuropsychological disorders or external influences. Posttraumatic stress disorder (PTSD) is one of the most closely related examples. During PTSD patients cannot be released from the memory of a traumatic event because it returns in the form of nightmares, recurrent recollections or flashbacks. In these cases the main question is why patients are unable to forget and what kind of process disables the suppression of unpleasant memories.
To understand this phenomenon we have to also understand the background of forgetting. Interference-based approaches emphasize the role of cue-target associations. Target is the memory we want to recall and cue is the key to the retrieval (e.g. a monogram or first letters in case of remembering a name). According to these theories, forgetting is caused by weakening the cue-target associations and recall success depends on the ratio of connection between the cue and the target and connections between cue and other targets [4,5,6]. However, we have to face with many phenomena and related results which can hardly be explained by the theories mentioned above, thus we have to assume the active contribution of control processes and presence of memory inhibition . This process occurring during accidental retrieval is similar to the motor response inhibition which is necessary to prevent and overwrite our reflexive actions with another behavior (e.g. when we experience an unpredictable event during driving and make a sudden steering in order to prevent carom breaking off our routine actions). To investigate the effect of a similar inhibition process in context of retrieval on the later recall performance and how does it lead to forgetting Anderson and Green developed the Think/No think task, which is a combination of the Go/No go (Figure 2, left side) and retrieval practice paradigms [2,3]. During this task participants have to study cue-target pairs (e.g. steam-train) then they have to retrieve the target with the help of the cue during many cycles. But, in certain cases (here: the cue presented with red color) they have to prevent the retrieval and do not think about related target (No think pairs) as can be seen in Figure 1. Their results suggest that this instruction leads to significant decrease in the later recall of No think items. These items are associated with an active inhibition during the practice and we can see this effect on the later recall.
Figure 1: Think/No think task
According to the literature of control processes we distinguish between automated and controlled inhibition depending on the associative learning related to stimulus-response connections. . In case of Go/No go task (Figure 2, left side) response inhibition is automated because we know the appropriate response at the moment when the stimulus appears. In contrast, there is another widely used executive function paradigm, the Stop Signal task (Figure 2, right side), which involves controlled inhibition as the result of inconsistent stimulus-response mapping. In this case stopping instruction follows the presentation of stimulus with a delay, in the form of a high-pitched sound. In case of Think/No think task participants never start to retrieve the No think elements, therefore, the response can be associatively learned and inhibition can be automated. To investigate this phenomenon during a situation involving controlled, effortful inhibition we combined the retrieval practice paradigm with Stop Signal task instead of Go/No go. In our task participants receive the Stop Thinking (No think analogue) instruction with a delay in the form of high-pitched sound instead of getting it immediately with the cue. We suppose that suppression can be also observed in this case and we will experience forgetting despite participants have to stop an already started retrieval which does not allow a consistent stimulus-response mapping in comparison with Go/No go-based Think/No think task. Due to the characteristic of Stop Signal paradigm, we count on an increased amount accidental retrieval despite the Stop instruction which opens an opportunity to investigate the effect of false alarms on later recall.
Figure 2: Go/No go and Stop Signal tasks
In our study at first participants have to learn fifty Swahili-Hungarian pairs of words presented one by one with an interstimulus interval of 500 msec, during five study cycles. Each word pair is shown for five seconds. This is followed by the Stop Thinking phase during 16 cycles. 40 pairs out of 50 words are presented here. In this phase a Swahili word appears then participants have to retrieve the associated Hungarian word and indicate the successful retrieval with a key press. However, if they hear a sound after the presentation of the cue they have to interrupt retrieval, do not think further on the Hungarian word, avoid pressing the button and prepare themselves for the next presentation. They get signal on ten words (No think pairs) out of 40, thus participants have to retrieve 75% of the words (Go pairs) during practice. One half of participants gets the signal with 250 msec onset (short onset group) and the other half gets it with 1000 msec onset (long onset group). That phase is followed by a five minutes long distractor task, and then the final recall. All of 50 Swahili cues are presented here and participants have to recall the associated Hungarian targets. If the retrieval is successful they have to indicate it with a key press and then type in the Hungarian target. The task is demonstrated on Figure 3.
Figure 3: Stop Thinking task
We calculated the hit rate during the retrieval practice (number of responses on Go stimuli / 16 cycles) and the false alarm rate for the Stop stimuli (number of responses on Stop stimuli / 16 cycles). Median reaction time for the correct responses and recall success rate was measured during the final recall.
Fifty healthy undergraduate students took part in the experiment for course credits from ELTE. Participants were randomly grouped in the short and long onset conditions, so these conditions contain 25-25 persons at the mean age of 20-22 years.
We did not observe any significant difference by the overall performance (condition independent recall rate, reaction time) and by the miss rates (1-hit) between short and long onset groups, in contrast, increased amount of false alarm was shown by the long onset group as can be seen in Figure 4. This result suggests that our task works as a reliable Stop Signal paradigm, because the onset of sound is a key indicator for both the difficulty and efficiency of inhibition and there is no additional difference between groups.
Figure 4: Error rates in Stop Thinking task separated by onset
That was followed by the investigation of efficiency and speed of retrieval. We compared the performance on Go, Stop and Baseline pairs. We can observe significant difference only between recall of Go and Baseline elements, the performance on Stop elements take place between them and does not show any significant difference from any stimulus group (Figure 5, left side). According to the reaction times we can see a strong priming effect (faster reaction time compared to the Baseline) on all of the stimuli appearing in Stop Thinking task independently from the condition (presence of a stop instruction), however, this effect is moderately weaker in the short onset group on the Stop word (Figure 5, right side). To summarize, Practice cycles result a strong priming effect independently of the instruction both by Stop and Go pairs and seemingly does not lead to forgetting despite the stop instruction.
Figure 5: Results of the final recall separated by condition and onset
We can ask the question why we do not experience neither suppression effect nor inhibition in case of Stop pairs. This can be explained by the increased amount of false alarms. This is negligible by the Think/No think task, however, in the Stop Thinking task it could be determinative on final recall and forgetting. In order to get a clearer result we checked how is the probability of successfully recalling a single item determined by the false alarm rate and hit rate during the practice phase. Logistic regression analysis reveals different connections (beta and OR) by Stop (Figure 6, lower row) and Go words (Figure 6, upper row).
Figure 6: Probability of successful recall as the function of hit rate and false alarms
Based on above we were able to make distinction between perfectly inhibited (16/16 cycles, STOP100 condition) and partly unsuccessfully inhibited Stop elements and to control this we also separated totally unsuccessfully practiced GO elements (0/16 cycles, GO0 condition) from the other practiced elements. Both conditions contain elements only with zero response (key press). In the following we investigated the probability of later successful recall depending on the described stimulus grouping. In the GO0 condition none of the participants gave any correct responses to the Swahili cues (0% performance), thus this stimulus group cannot be seen on Figure 7. Our results indicate perfect inhibition (STOP100 column) leads to significant suppression tending to the Baseline level by the short onset group and below the Baseline by the long onset group, additionally, partly unsuccessful inhibition (STOP column) results facilitation above or similar to the Go elements which were at least once retrieved during the practice phase (Figure 7, left side). Analysis of the reaction time reveals that the priming effect for the perfectly inhibited pairs is significantly lower in comparison with Go and other Stop elements (Figure 7, left side). Therefore, we can conclude that only perfect inhibition leads to suppression, however when participant accidentally retrieves the target despite the stop instruction, accessibility and later recall probability of this element increases like it were practiced during the 16 cycles.
Figure 7: The effect of successful and partly unsuccessful inhibition on later recall
In previous approaches of the memory inhibition suppression effect of automatic and controlled inhibition was not separated according to our knowledge, thus the consequences of stopping an already started retrieval was not examined earlier. Therefore, our results are suitable both to enlarge the conceptual framework of the memory inhibition and provide harmony with the literature of control functions, and makes the paradigms described in the introduction more widely applicable for everyday phenomena.
Stop Thinking paradigm could also be a milestone in understanding the maladaptive memory phenomena experienced in case of posttraumatic stress disorder. The basis of the above mentioned flashback effects is the result of unsuccessful attempt for inhibition to retrieve an unwanted memory, which – in accordance with our results – increases the probability of the accidental retrieval. In addition, the described consequences could help us investigate the symptoms of situation-specific psychogenic amnesia, in which patients are unable to retrieve the events of a time period surrounding a traumatic event. Our results suggest that perfect inhibition leads to enormous forgetting and confirm that an increased operation of inhibition processes may be accountable for this phenomenon.
Kárpáti J., Markója Á., Kónya A., Király I. (2016): A tér-idői emlékezeti integráció hátterében álló kontrollfolyamatok vizsgálata gyermekeknél (in Hungarian). In: Múlt és jelen összeér, XXV. Scientific Assembly of the Hungarian Psychological Association; Abstracts, 2-4 June, 2016, Poster 13, p. 348
Markója Á., Kárpáti J., Kónya A. (2016): A Stroop paradigma és a Trail making teszt: a végrehajtó funkciók vizsgálata (in Hungarian). In: Múlt és jelen összeér, XXV. Scientific Assembly of the Hungarian Psychological Association; Abstracts, 2-4 June, 2016, Poster 06, p. 334
Markója Á., Szőllősi Á., Racsmány M. (2017): Előhívás, felejtés és gátlás (in Hungarian). In: Személyes tér – közös világ, XXVI. Scientific Assembly of the Hungarian Psychological Association; Abstracts, 1-3 July, 2017, Symposium 39, p. 163
1. Anderson, M. C., Bjork, R. A., & Bjork, E. L. (1994). Remembering can cause forgetting: retrieval dynamics in long-term memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 20(5), 1063.
2. Anderson, M. C., & Green, C. (2001). Suppressing unwanted memories by executive control. Nature, 410(6826), 366-369.
3. Anderson, M. C., & Levy, B. J. (2009). Suppressing unwanted memories. Current Directions in Psychological Science, 18(4), 189-194.
4. McGeoch, J. A. (1942). The Psychology of Human Learning, an Introduction New York: Longman
5. Melton, A. W., & Irwin, J. M. (1940). The influence of degree of interpolated learning on retroactive inhibition and the overt transfer of specific responses. The American Journal of Psychology, 53(2), 173-203.
6. Rundus, D. (1973). Negative effects of using list items as recall cues. Journal of Verbal Learning and Verbal Behavior, 12(1), 43-50.
7. Verbruggen, F., & Logan, G. D. (2008). Automatic and controlled response inhibition: associative learning in the go/no-go and stop-signal paradigms. Journal of Experimental Psychology: General, 137(4), 649.
During Go/No go task participants have to respond with key press on a set of a stimuli (Go stimuli, e.g. blue shapes) and avoid giving response to an other set (No go stimuli, e.g. red shapes). When they have to response on the No go stimuli these answers will be slower due to the inhibition.
Examination of the effect of the continuous retrieval on the later recall performance.
Baseline elements: 10 items not appearing during the practice cycles, they were only presented during the study phase and retrieved during the final recall.