THE EFFECTS OF TYPE OF RETRIEVAL ON PREDICTED AND ACTUAL MEMORY PERFORMANCE FOR
AN EPISODIC LIE-GENERATION PARADIGM
A Master’s Thesis
by
GAMZE NUR EROĞLU
The Department of Psychology İhsan Doğramacı Bilkent University
Ankara August 2020 GA MZE N UR ER OĞ LU EP ISOD IC LY IN G, MEMO R Y, A ND META MEMORY B il ke nt Univer sit y 20 20
THE EFFECTS OF TYPE OF RETRIEVAL ON PREDICTED AND ACTUAL MEMORY PERFORMANCE FOR
AN EPISODIC LIE-GENERATION PARADIGM
The Graduate School of Economics and Social Sciences of
İhsan Doğramacı Bilkent University
by
GAMZE NUR EROĞLU
In Partial Fulfillment of the Requirements for the Degree of MASTER OF ART IN PSYCHOLOGY
THE DEPARTMENT OF PSYCHOLOGY
IHSAN DOĞRAMACI BILKENT UNIVERSITY ANKARA
ABSTRACT
THE EFFECTS OF TYPE OF RETRIEVAL ON PREDICTED AND ACTUAL MEMORY PERFORMANCE FOR
AN EPISODIC LIE-GENERATION PARADIGM
Eroğlu, Gamze Nur M.A., Department of Psychology Supervisor: Asst. Prof. Dr. Miri Besken
August 2020
Intentional generation of lies is a widely studied topic that has attracted attention over the last two decades. However, the memory for one’s intentional lies has not been studied thoroughly. Some studies suggest that deceptive answers intrude into one’s memory as false memories, however, most of the time, the results come from different types of paradigms with different types of retrieval tests. Theoretically, one factor that can potentially change memory of one’s lies might be the type of retrieval that they have to engage in. The current study investigated how using different types of retrieval such as free-recall, cued-recall as well as source and destination
recognition may change both actual and predicted memory for lies and the truth. In a set of 3 experiments, participants were asked to tell the truth or tell a lie in the
questions through cued-recall (Experiment 1), free-recall (Experiment 2) or source and destination recognition (Experiment 3). Experiments 1 and 2 showed that according to response latencies, lying was more difficult than telling the truth. This difficulty was not reflected in participants’ predictions, truthful and deceptive
answers were predicted to be remembered equally well. Actual memory performance differed across experiments: truthful answers were remembered more in Experiment 1, and deceptive answers were remembered more in Experiment 2. The results imply that the type of retrieval may change the pattern of actual memory performance between truth and lies, even though this is not reflected in memory predictions during encoding. Experiment 3 investigated whether lying would be influenced by the contextual information, associated with retrieval type by using a source and destination retrieval task. Participants were asked to tell truthful or deceptive information to the people on the screen, or receive truthful or deceptive information from the people on the screen. Results revealed that participants were able to recognize the faces from whom they received information more than the faces they told information to, regardless of the accuracy of the information. The results are discussed with processing fluency hypothesis and source monitoring framework.
Keywords: Cued-recall, Destination Memory, Free-recall, Lying, Metamemory, Source Memory
ÖZET
ANISAL YALAN ÜRETME PARADİGMASI İÇİN GERİ ÇAĞIRMA TİPLERİNİN TAHMİN EDİLEN VE ASIL BELLEĞE OLAN ETKİLERİ
Eroğlu, Gamze Nur Yüksek lisans, Psikoloji
Tez Danışmanı: Dr. Öğr. Üyesi Miri Besken
Ağustos 2020
Kasıtlı yalan üretme, son yirmi yılda ilgi çekmiş ve yaygın olarak araştırılan bir konudur. Buna karşın, kişinin kasıtlı söylediği yalanlara dair belleğinin nasıl etkilendiği derinlemesine incelenmemiştir. Bu konudaki bazı çalışmalar yanıltıcı cevapların sahte anı şeklinde belleğe girdiğini gösterse de sonuçlar genellikle farklı geri çağırma yöntemi kullanan farklı paradigmalardan elde edilmiştir. Teorik açıdan, kişinin belleğini etkileme potansiyeline sahip olan faktörlerden biri kullanılan geri çağırma yöntemidir. Bu araştırma, serbest ve ipuçlu hatırlama ve hatta hedef ve kaynak belleği tanıma gibi farklı geri çağırma yöntemlerinin belleği ve üstbelleği yalan ve doğru cevaplar açısından nasıl etkilediğini incelemektedir. Üç deneyde de katılımcılar kodlama aşamasında soruları doğru ve yalan söyleyerek cevapladıktan sonra cevaplarını daha sonra hatırlayacaklarından ne kadar emin olduklarını
2’de yalan söylemenin işlenme açısından daha zor olduğu cevap verme sürelerine göre gösterilmiştir. Fakat bu zorluklar üst belleğe yansımamış, doğru ve yanıltıcı cevapların eşit şekilde hatırlanacağı tahmin edilmiştir. Asıl bellek performansları ise deneyler arasında farklılık göstermiştir: Deney 1’de doğru cevaplar, Deney 2’de ise yanıltıcı cevaplar daha fazla hatırlanmıştır. Sonuçlar, geri çağırma yönteminin doğru ve yalan arasındaki bellek performansı ilişkisini etkilemesine rağmen bellek
tahminlerinde bir farklılık oluşmadığını göstermektedir. Deney 3, geri çağırma tipine bağlı olan bağlamsal bilgilerin yalan söyleme durumundan etkilenip
etkilenmeyeceğini hedef ve kaynak tanıma belleği testi ile incelemektedir. Katılımcılardan ekranda gördükleri kişiye doğru veya yalan bilgi vermeleri veya ekranda gördükleri kişiden doğru veya yalan bilgi öğrenmeleri istenmiştir. Sonuçlar, bilginin doğruluk değeri fark etmeksizin katılımcıların bilgi aldıkları kişileri, bilgi verdikleri kişilerden daha iyi tespit ettiklerini göstermiştir. Sonuçlar, işlemsel akıcılık hipotezi ve kaynak izleme çerçevesi ile tartışılmaktadır.
Anahtar Kelimeler: Hedef Belleği, İpuçlu Hatırlama, Kaynak Belleği, Serbest Hatırlama, Üst Bellek, Yalan
ACKNOWLEDGMENTS
I cannot express enough gratitude to my advisor and more, Asst. Prof. Dr. Miri Besken. Both emotionally and academically her support and guidance to me were immensely valuable. Her vast knowledge, eagerness to work, keen interest to her students and sense of humor made it a great privilege to work with her.
I would like to thank Asst. Prof. Dr. Jedediah Wilfred Papas Allen and Asst. Prof.
Dr. Aslı Kılıç Özhan for accepting to be a member of the examining committee of
this exhaustively long thesis, not sure whether they knew what they got themselves
pulled into this by accepting it, but I owe special thanks to both of them.
I owe my deepest gratitude to my family: my mother Nuran, father Erdal, and sister Ayşegül Beyza. Their everlasting love and belief in me served as the greatest
motivation in my life. I consider myself as the luckiest for having them in my life. I
will also include my partner in this paragraph, Can Öztürk, who has a unique place in
my heart and became my family after many years. To me, his presence in my life is
the dawn that illuminates the day after the darkest hour.
I also appreciate the members of Memory Lab for their assistance in data collection. I would like to thank my thesis-date, Ezgi Melisa Yüksel. Her radiant energy, valuable feedback, and support made the process of writing became pleasurable to the maximum. Last, but not least, I thank Gizem Filiz, Ecem Eylül Ardıç, and Elif
TABLE OF CONTENTS
ABSTRACT ... iv
ÖZET... vi
ACKNOWLEDGMENTS ... viii
TABLE OF CONTENTS ... ix
LIST OF TABLES ... xiii
LIST OF FIGURES ... xiv
CHAPTER I ... 1
INTRODUCTION ... 1
1.1 Lying and Memory ... 2
1.1.1 Forced and Voluntary Confabulation ... 3
1.1.2 Deliberate Lying ... 4
1.2 Types of Lies ... 5
1.2.1 Elaboration ... 5
1.2.2 False Denial ... 6
1.3 Types of Retrieval ... 7
1.3.1 Destination and Source as Types of Recognition Tests ... 9
1.3.1.1 Testing Methods ... 10
1.3.1.2 General Findings ... 13
1.3.1.3 Deception in Source and Destination Memory ... 14
1.4.1 Behavioral Evidence ... 19
1.4.2 Neurocognitive Evidence ... 21
1.4.3 Contrary Evidence ... 21
1.5 Metamemory and Judgments of Learning ... 23
1.5.1 Lying and Metamemory ... 25
1.6 Aims of the Current Study ... 27
CHAPTER II ... 30 EXPERIMENT 1 ... 30 2.1 Participants ... 30 2.2 Materials ... 31 2.3 Design ... 31 2.4 Procedure ... 32 2.5 Results ... 36 2.5.1 Procedure Compliance... 38 2.5.2 Response Latencies ... 38
2.5.3 Predicted Memory Performance, JOLs ... 39
2.5.4 Cued-recall Memory Performance (Proportional) ... 40
2.5.5 Truth-check Memory (Proportional) ... 41
2.6 Discussion ... 42
CHAPTER III ... 45
EXPERIMENT 2 ... 45
3.1 Participants ... 45
3.5 Results ... 46
3.5.1 Procedure Compliance... 48
3.5.2 Response Latencies ... 48
3.5.3 Predicted Memory Performance, JOLs ... 49
3.5.4 Free-Recall Memory Performance (Proportional) ... 49
3.5.5 Truth-check Memory (Proportional) ... 51
3.6 Discussion ... 52 CHAPTER IV ... 55 EXPERIMENT 3 ... 55 4.1 Participants ... 55 4.2 Materials ... 56 4.3 Design ... 57 4.4 Procedure ... 58 4.5 Results ... 62 4.5.1 Procedure Compliance... 64 4.5.1.1 Source Memory ... 64 4.5.1.2 Destination Memory ... 64 4.5.2 Key-Press Latencies ... 65
4.5.2.1 First Key-press Latency ... 65
4.5.2.2 Response Completion Latency ... 66
4.5.3 Predicted Memory for Items ... 67
4.5.4 Predicted Memory for Faces ... 69
4.5.5 Corrected Item Memory ... 72
4.5.6 Corrected Face Memory ... 74
4.5.8 Truth-check Memory ... 76
4.5.8.1 Further Analysis on Type of Mistakes in Truth-Check Phase ... 78
4.5.9 Tendency to Lie ... 80
4.6 Discussion ... 82
CHAPTER V ... 88
GENERAL DISCUSSION... 88
5.1 Reminder of the Aims and Discussion of the Results from a Theoretical Perspective ... 88
5.2 Limitations and Future Directions ... 95
5.3 Theoretical and Applied Implications ... 99
REFERENCES ... 101
APPENDICES ... 109
A. LIST OF QUESTIONS IN EXPERIMENT 1 ... 109
B. EXAMPLARY PICTURES IN EXPERIMENT 1 AND 2 ... 114
C. LIST OF QUESTIONS IN EXPERIMENT 2 ... 115
D. LIST OF QUESTIONS AND ANSWERS IN EXPERIMENT 3 ... 118
D.1. VERSION 1 ... 118
LIST OF TABLES
Table 1. Means (Standard Error of the Means) for Median First Key-press and Response Completion Latencies, Mean JOLs, Proportional Cued-recall and Truth-check Memory Performance for Truth and Lie Trials in Experiment 1 ... 37 Table 2. Means (Standard Error of the Means) for Median First Key-press and
Response Completion Latencies, Mean JOLs, Proportional Free-recall and Truth-check Memory Performance for Truth and Lie Trials in Experiment 2 ... 47 Table 3. Means (Standard Error of the Means) for Key-press Latency, Predicted and Corrected Memory Performances for Items and Faces, and Proportional Truth-check Memory Performances for Destination and Source Memory Conditions and Truth and Lie Trials in Experiment 3... 63 Table 4. Means (Standard Error of the Means) for Item and Face Memory
Performance in Destination and Source Conditions in Truth and Lie Trials ... 72 Table 5. Number of Cases in Mistake Types across Destination and Source Memory Conditions in Lie and Truth Trials ... 80
LIST OF FIGURES
Figure 2.1 Schematic display of the encoding phase in Experiment 1 and 2... 34
Figure 2.2 Schematic display of the memory test phase in Experiment 1 and truth-check phase in Experiment 1 and 2 ... 36
Figure 4.1 Schematic display of the encoding phase in Experiment 3 ... 59
Figure 4.2 Schematic display of the memory test phase in Experiment 3 ... 60
Figure 4.3 Means of Medians for First Key-press Latency in Experiment 3 ... 65
Figure 4.4 Means of (A) Predicted and (B) Corrected Item Memory Performance in Experiment 3 ... 68
Figure 4.5 Means of (A) Predicted and (B) Corrected Face Memory Performance in Experiment 3 ... 71
CHAPTER I
INTRODUCTION
How often do you think that people lie? Think about the last incident that you lied about. One can lie in order not to offend others’ feelings, to maintain personal interests which are typically considered as acceptable situations to lie due to their prosocial function (Serota & Levine, 2015; Sweetser, 1987). Yet, there are also situations in which one can also lie about their feelings for a person, by calling in sick while feeling well. These are typically considered as unacceptable situations to lie and they serve self-protective function to the liar (Serota & Levine, 2015; Xu, Bao, Fu, Talwar, & Lee, 2010). Although prosocial lying is more frequent than self-protective lies (DePaulo & Bell, 1996) and they have different functions in social life, when approached broadly we can claim that lying is a very common
phenomenon. DePaulo, Kashy, Kirkendol, Wyer, and Epstein (1996) demonstrated that people in the community engage in such lying incidents at least once a day, and this number increases to at least twice a day for college students.
The current study aims to investigate the actual and predicted memory performance for episodic lies in relation to the type of retrieval that the participants have to engage in. The episodic lie-generation term is used in the current study in terms of
“fabricating an incorrect response by changing the previously seen detail.” The introduction chapter will start with a discussion of how the concept of lying is studied in adult memory literature and what are the general results. Next, I will review how different types of lies influence the pattern of results. Then, types of retrieval will be introduced as one of the factors that produce different results in many well-known paradigms and an example for how retrieval types influence false memory differentially will be explained. In types of retrieval section, destination and source recognition tests will be introduced as one of the methods to measure
recognition memory and how the act of lying is used by destination and source memory as well as their general testing methods and general findings will be
explained. Following this, theoretical frameworks for the effect of lying on memory will be presented as effortful processing theory and source monitoring framework. After the characteristics of effortful processes are outlined and the relevance of source monitoring framework is explained, evidence that lying is processed more effortfully than telling the truth will be provided. Towards the end of introduction, the focus will shift to the importance of memory predictions and how lying
influences memory and metamemory will be discussed. Lastly, the aims of the current study will be stated along with our research questions.
of both detection (DePaulo et al., 2003; Granhag & Hartwig, 2008; Vrij, Fisher, & Blank, 2017; Vrij, Granhag, Mann, & Leal, 2011; Walczyk, Igou, Dixon, & Tcholakian, 2013) and production of lies (DePaulo et al., 1996; DePaulo & Kashy, 1998; Serota, Levine, & Boster, 2010; Verigin, Meijer, Bogaard, & Vrij, 2019). However, how well a piece of information is remembered after telling a lie versus after telling the truth has not been investigated so thoroughly and is a relatively new topic of interest. The data for the memory of lies is mostly investigated through studies of forced and voluntary confabulation. The foregoing sections will focus on the effects of voluntary and forced confabulation on memory, followed by the few studies that investigate the effects of deliberate lying on memory.
1.1.1 Forced and Voluntary Confabulation
In daily life, one can be forced to choose a response among a limited number of response options which may or may not be suitable to the person, or none of them might be the actual answer. Especially in forensic cases and police stations, one can be forced to provide a testimony up until a point that the person confabulates some information that never happened in order to terminate the interrogation procedure. It has been shown that forced confabulation is associated with false memories (Ackil & Zaragoza, 1998, 2011; Zaragoza, Payment, Ackil, Drivdahl, & Beck, 2001). In these studies, participants were asked to witness an event followed by questions both related and unrelated to the event they witnessed and were asked to answer all questions. One week later, a memory test was conducted, which generally revealed that participants remembered the information they were forced to confabulate as if it happened in the actual event they witnessed. In other words, they were not able to distinguish between the original events and the details that they confabulated about.
In another study, the forced and voluntary confabulation conditions were compared in terms of false memory rates (Pezdek, Sperry, & Owens, 2007). Similar to the studies described above, participants were presented with an event and asked to answer both related and unrelated questions. Differently from the others, participants in the voluntary confabulation condition had “I do not know” option that they could choose for the questions they did not know the correct option, whereas the ones in the forced confabulation condition did not have “I do not know” option and were asked to make a guess. One week later, both groups were provided with “I do not know” option on a memory test with the same questions. The results showed that the voluntary confabulation condition had higher false memory rates, which suggests that the memory is even more vulnerable to alterations during voluntary
confabulations than forced confabulations.
1.1.2 Deliberate Lying
The abovementioned studies utilized slides or videos of an event and presented them on a screen for participants to witness, followed by a confabulation phase, altering details of the episodic event during the testing phase. Despite the fact that
confabulation is not conceptually the same thing as a deliberate attempt at lying, the act of intentional lying during encoding may also alter memory traces. Studies that investigated the effects of deliberate lying on memory are not as vast as the ones that investigated eyewitness memory and confabulation (see Pickel, 2004; Polage, 2004; Sauerland, Schell-Leugers, & Sagana, 2015; Vieira & Lane, 2013). How telling a lie upon instruction alters the memory and how different types of lies influence the
1.2 Types of Lies
In the literature, different procedures were used in the lie-generation manipulations. The procedures can be categorized as false denial (i.e., participants are asked to deny having seen the item or committing the action where in reality, they actually saw or committed) and fabrication (i.e., participants are instructed to elaborate on the details of objects or events they have not been exposed to, or provide alternate details
contrary to the real objects or events they were exposed to). How false denial and fabrication influence memory are explained further in the following sections.
1.2.1 Elaboration
In Pickel (2004), participants were shown a video of a robbery and immediately after the video presentation, they were interrogated as a witness in three different
conditions: fabricate information about the clerk, fabricate information about the thief, answer all questions honestly. There was also a control group in which the participants were not interrogated about the event at all. One week later, all
participants were asked to provide information about the physical appearances of the thief and the clerk based on what they witnessed in the video, which means the participants who were interrogated in any of the three conditions were asked to disregard the information they provided during the interrogation.
The highest false memory rate was observed for the participants in the fabrication condition. More specifically, if they fabricated about the clerk, they falsely
remembered their fabricated information about the clerk during the part where they were supposed to answer truthfully based on the video; and if they fabricated about the thief, their false memory rate was higher for the thief. Thus, we can deduce that
providing false information about a person leads to falsely remembering them as if this was the actual information, but this is exclusive to the target of fabrication, fabricating information about the clerk did not influence false memories for the thief and vice versa. Thus, when participants lied about a person, the act of lying overrode their memory for the actual event, they remembered their deceptive answers for that specific target as the correct information which was presented in the video.
1.2.2 False Denial
In another study, different types of lies, false denial and elaboration, were compared in terms of their effects on memory (Vieira & Lane, 2013). The results revealed that memory performance was higher for the participants who elaborated on the details about the items that they did not see, compared to the participants who falsely denied having seen the items that they actually saw. Even though participants were
instructed to lie in both conditions, it was shown that the type of lying produced differences in memory performance. The authors claimed that elaboration required more cognitive effort than simply denying by saying “no” and these difficulties while lying might have favored the elaboration condition in remembering the previously told details. Other studies have also shown that denial and elaboration are not similar in terms of the amount of cognitive resources they deplete (Otgaar & Baker, 2018; Walczyk, Harris, Duck, & Mulay, 2014).
Otgaar and Baker (2018) claimed in an extended review that different types of lying may produce different effects on memory, depending on the degree of cognitive
other words. For example, in Romeo, Otgaar, Smeets, Landstrom, and Boerboom (2019) virtual reality (VR) experience of a scene that depicted airplane crush was provided to participants to increase their interaction with the traumatizing material both auditorily and visually. Participants either told the truth or were directed to falsely deny by reiterating the question. Results showed that memory performance was lower in the directed false denial condition than the truth-telling condition both in terms of interview and the actual details in the VR experience. In contrast, elaboration increases the memory performance for the fabricated details, which is commission error, or a false alarm, as explained in the section above when participants falsely reported details that they fabricated as the correct information (Pickel, 2004) Based on this information, we can claim that elaborating on details that never happened distorts memory for the actual information by overriding them, and the elaborated details are generally remembered as if they were the correct information.
1.3 Types of Retrieval
Regardless of the nature of the lying task, there is good reason to think that different types of retrieval could also produce patterns of different memory performance. For example, in free-recall, participants are expected to remember as many items as possible in the allocated time without any external cues. This method is informative about the internal organizational structure of the memory, because it allows
participants to use self-initiated retrieval strategies (Cleary, 2018). In cued-recall, in contrast, participants are provided with external cues (e.g., category name, pairs, visual information) to help their retrieval attempts. Most of the time, cued-recall
leads to higher memory performance than free-recall (see Reffel, 1997 for an exception).
There are cases in which cued- and free-recall produce different results. For instance, bizarreness effect, i.e., items presented in bizarre context leads to higher retrieval than items presented in common context, produces different pattern of results when tested with a free-recall than cued-recall (Riefer & Rouder, 1992). Riefer and Rouder (1992) investigated the bizarreness effect by using noun pairs. For example, chef-pickle pair could be presented within a common context, “The CHEF sliced the PICKLE”, or within a bizarre context, “The CHEF smoked the PICKLE”. In the free-recall test, participants were asked to remember the capitalized words followed by a cued-recall test in which the first capitalized noun was provided and participants were asked to remember the second capitalized noun. They reported that memory enhancing influence of bizarreness is valid only when measured with free-recall and no difference between bizarre and common contexts is observed with cued-recall.
False memory might also be sensitive to the retrieval method. Ackil and Zaragoza (2011) compared the false memory rates between forced confabulation and suggestions from the interviewer. Of relevance to the current discussion, they compared the type of retrieval test by using two different types of test: source recognition and narrative recall. The pattern of results was in the opposite direction between recognition and recall tests. That is, when participants were asked to recognize the source, external suggestions led to higher false memories than forced
condition. Thus, the assessment method used in the interviews, recognition or free-recall, produced different patterns of the results when forced confabulation and suggestions were compared.
1.3.1 Destination and Source as Types of Recognition Tests
Even though different types of retrieval methods (e.g., free-recall, cued-recall, or recognition) may lead to different pattern of memory performance, it is also possible to disentangle different types of recognition tests: destination and source. Lying is inseparable from social context as it requires at least one perpetrator and one
recipient. Thus, it is important to remember both to whom a lie is told or from whom deceptive information is received. Having this information might be just as important as remembering whether given/received information is truthful or deceptive.
In the memory literature, remembering the context of the information is just as valuable as remembering the item itself. When participants remember the content of the information, that is considered item memory. Remembering the context of the information in addition to the content takes different forms. Source memory refers to remembering when, where or from whom the previously learnt information is
acquired. A similar concept to source memory is destination memory, referring to the memory of whom the previously told information is given (Gopie & Macleod, 2009). The context of source and destination memory can refer to various contextual
characteristics, such as location, time, the medium of learning, the person that the information is learnt from or given to (Lindsay & Johnson, 1991).
As one of the aims of the current set of studies is to document how source and destination recognition for lies may compare, the following sections provide a brief review of the source and destination memory literature related to the testing methods and general findings. Afterwards, the use of source and destination memory in relation to deception literature will be discussed.
1.3.1.1 Testing Methods
Source memory is generally tested with detecting the source of the information after multiple learning contexts. The context could be slides, videos, texts, or interviews (Johnson et al., 1993). For instance, Vieira and Lane (2013) presented participants with various pictures of everyday items on a computer screen. Then, participants were asked to give verbal statement to the camera by either describing or denying seeing the item. In the source recognition test, participants reported whether they had studied the item or not and whether they described, denied or did not say anything about it while they were giving verbal statement. Results indicated that source memory for items that were described were higher than those which were denied.
As well as verbal statements, the source could also include written statements. In a study by Lane and Zaragoza (2007) after viewing the slides which depicted a robbery, participants were given several paragraphs about the event they witnessed but some portions of the details were not present in the slides. For the critical items presented in the paragraphs but were absent in the slides, participants either
Compared to the read-elaboration and the read-item-only conditions, generating details for critical items resulted in the highest correct source attribution rates (i.e., generated information was correctly attributed to paragraphs rather than the slides), however, it also resulted in the highest source misattribution rates (i.e., generated information was falsely attributed to slides).
The source recognition test could also be about detecting the person who told the specific information after learning various information from different people (Johnson, Hashtroudi, & Lindsay, 1993). Dodson, Holland, and Shimamura (1998) auditorily presented participants with words by two male and two female speakers. It was observed that participants were able to accurately detect the gender of the speaker for the items they failed to detect the accurate source, even when their attention was divided during a source memory test to remember a sequence of the numbers which was heard before each item.
Destination memory, on the other hand, is generally tested with a narrower scope, i.e., detecting the person to whom information is told after telling various
information to different people (Gopie & Macleod, 2009). For example, in a pioneer study in this area, 60 pieces of interesting information and 60 pictures of famous people were used. Participants in the destination condition were presented with the information first, followed by the pictures and were asked to tell the information out loud to the celebrity on the screen. Participants in the source condition were
presented with the pictures first, followed by the information and were instructed that they were learning the information from that celebrity. For the memory performance, they measured both item and associative memory by using a recognition test. For the
item memory, information (10 new and 10 previously seen old information) and faces (10 new and 10 old faces) were presented separately and participants indicated whether they saw the item or not. For the associative memory, 40 face-information pairs were presented simultaneously, but half of them were identical pairs, and the other half were mixed within the other 20 faces and pictures. Participants were asked to indicate whether that specific information was told to or received from the person on the screen. As a result, participants in the source memory condition had higher face recognition and associative memory performance than participants in the destination memory condition, whereas for item memory there was no difference between source and destination conditions.
In addition to the other studies that utilized celebrity pictures as source and
destination (El Haj, Omigie, & Samson, 2015; El Haj, Saloppe, & Nandrino, 2018), there are also studies with higher ecological validity. For instance, in Fischer, Schult, and Steffens (2015), a speed-dating set up was utilized. Participants in the destination condition were asked to complete personal information fragments (e.g., my
horoscope is …) and tell the person in front of them, while the ones in the source condition were listening to them. For face recognition, result from Gopie and Macleod (2009) was replicated, source condition had higher performance than destination condition. However, interestingly, item recognition was higher in destination than source condition. In another study (Lindner, Drouin, Tanguay, Stamenova, & Davidson, 2015), participants in the destination condition, read the given information to two different experimenters whereas the ones in the source
and if not new, which experimenter the information was associated with. Lindner et al. (2015) reported that for item memory, destination condition was superior than source condition but for recognizing the person, destination and source conditions were comparable. The general finding that destination memory is lower than source memory was observed when participants were asked to say the sentences to
experimenters from memory.
1.3.1.2 General Findings
So far, the studies comparing destination and source memory reported that source recognition produces higher memory performance than destination recognition (Fischer et al., 2015; Gopie & MacLeod, 2009; see for an exception Lindner et al., 2015). There are two possibilities as to why source memory outperforms destination memory. The first possibility is that individuals are focused on themselves during transmission of information in destination procedure, which makes it more difficult to remember to whom the information is told (Gopie & Macleod, 2009). Based on this hypothesis, Gopie and Macleod (2009) manipulated self-focus across
experiments. In high self-focus group, participants were either told or received personal information, and superiority of source over destination remained the same. In low self-focus group, participants said the name of the celebrity first, and then either told or received general information. Contrary to the general finding, destination memory produced higher face recognition rate than source memory. Therefore, Gopie and Macleod (2009) claimed that higher self-focus during communication of information to others deteriorates destination memory performance.
The second possibility for lower memory for destination than source memory is the trade-off hypothesis (Fischer et al., 2015). Fischer et al. (2015) claimed that
participants encode contextual information (e.g., where, when) automatically in source memory tasks, because this contextual information might benefit participants in remembering “from whom” information. In contrast, participants are involved in either fabrication or transmission of the information for destination memory tasks, which might deplete the cognitive resources, leaving very few resources for encoding “to whom” information. Based on this hypothesis, Fischer et al. (2015) compared context and content memory. In context memory, they measured participants’ ability to recognize the person that either a piece of information was told to or received from. In line with the general finding, they found that source memory was higher than destination memory. In content memory, they measured participants’ ability in both conditions to recognize the information itself. Contrary to the general finding, they reported that the information that was either told to or received from others were remembered higher in destination than source condition, which supports the trade-off hypothesis about the opposing impact between content and context in destination and source memory. In source memory, the context information, the face information is remembered more, whereas in destination memory, the content information, the item information is remembered more.
1.3.1.3 Deception in Source and Destination Memory
One of the most critical situations in which remembering to whom a piece of information is told or from whom a piece of information is received is when the
Zaragoza (2007) focused on generation of did-not-witnessed details which was accompanied by forced confabulation rather than intentional lying (see Section 1.3.1.1). They showed that the mnemonic effect of generation comes with a cost: generating details leads to higher correct source recognition (hits) but also higher source misattribution (false alarms). Another study by Vieira and Lane (2013) compared different types of deliberate lies: false denial and elaboration. They reported that source memory was higher in elaboration than false denial (see Section 1.2.2).
To our knowledge, there is only one study that incorporated deception in destination memory (El Haj et al., 2018). In this experiment, participants were presented with 12 truthful general pieces of information (e.g., Rome is the capital of Italy) and 12 deceptive pieces of information (e.g., London is the capital of France) and they were asked to tell the given information to different celebrity pictures out loud.
Additionally, their tendency to lie in daily life was measured with the Impression Management Subscale of Paulhus Deception Scale (Paulhus, 1998). In the testing phase, participants were presented with pairs of information and celebrity pictures, in which half of them were the correct pair and the other half were mixed, and they were instructed to say “yes” if the given information was previously told to that celebrity and “no” if the given information was told to somebody else.
They reported that participants were able to detect the celebrities that they previously lied to more than the ones that they told the truth. Interestingly, this result was
whereas no significant difference was observed for participants who were categorized as having high tendency.
In addition to investigating whether we would remember the people that we told a lie previously more than the people we told the truth or not, it is also important to see whether we would detect the people who told us a lie among other sources more than the ones who told us truthful information. This ability, if we have, would be
beneficial in order to assess the trustworthiness of the source, both in social and forensic contexts. A handful of studies reported above that used deception paradigm in source recognition tests focused on different aspect of source than El Haj et al. (2018). That is how well a person can correctly identify the learning medium that the information was acquired at: generated versus read (Lane & Zaragoza, 2007) or denied versus elaborated (Vieira & Lane, 2013). Thus, it is important to see how the results reported by El Haj et al. (2018) unfolds for source recognition for detecting from whom the truthful or deceptive information was learned. In doing so,
controlling whether participants can correctly detect the accuracy of the given information (truth or lie) or not would also be an important aspect to consider.
Additionally, since this is the only study that uses deception in measuring destination memory, replicating these results with other materials than general information and celebrity pictures is also necessary.
1.4 Theoretical Framework for Lying and Memory
1993). I will briefly explain their reasoning and then examine how the act of lying maps onto their reasoning by providing evidence from behavioral and neurocognitive findings.
Hasher and Zacks (1979) claimed that processes involved in encoding can be conceptualized as being on a continuum in which automatic processes stand at one end and effortful processes stand at the other end. Automatic processes were characterized as follows: (1) they do not need deliberate intent to be processed and cannot be turned off when intended to do so. (2) They deplete very little cognitive resources. (3) They do not further benefit from practice but effortful processes can be automatized with practice. (4) Automatically encoded information is still accessible to retrieval. (p. 359-60). Examples for automatic processes may include reading or encoding spatial-temporal information and frequency of occurrence. We might effortfully pay attention to the spatial location of an object but we would encode the spatial information even in the absence of awareness and attempt to encode.
Characteristics of effortful processes can be listed as the following: (1) they need deliberate intent to be processed. (2) They deplete a great deal of cognitive resources. (3) They benefit from practice. (4) Individual differences might commonly be
observed (p. 362-3). Problem solving, elaboration, organization, and rehearsal can be given as examples of effortful processes since they are processed upon intentional instructions and can be improved with practice and feedback from results.
The source monitoring framework suggests a similar reasoning as effortful
discriminate the origins, i.e., source, of the memory there are competing cues: external and internal. The former is related to “discriminating memories of
statements that were made by person A from those made by person B”, whereas the latter is related to “discriminating memories of what one thought from memories of what one said” (Johnson et al., 1993, p. 4). External and internal cues differ in terms of their characteristics, namely external cues are assumed to have more perceptual, contextual, or affective references to the event itself, whereas internal cues are assumed to have more references regarding cognitive operations such as elaboration, imagination, or rehearsal. Efficacy of accurate source identification rely on the availability of the such cue characteristics in the memory traces and their uniqueness for different sources (p. 6). Thus, when internally generated events such as fabricated information have similar characteristics as externally experienced events, they might be treated as experienced events, which leads to source misattribution. The source monitoring framework was used as an explanation for false memories accompanied by lying in many studies (Otgaar & Baker, 2018; Vieira & Lane, 2013).
Based on the arguments given above, if lying is (1) an intentionally initiated process, (2) requires more cognitive resources than telling the truth, and (3) benefits from practice, we can claim that lying requires more effortful processing although both telling the truth and a lie are effortful processes, i.e., lying is placed closer to the effortful processing end of the continuum than telling the truth, and therefore lying potentially will be remembered more as more traces in the memory might lead source confusion during retrieval.
The behavioral and neurocognitive evidence for the premise that lying is processed more effortfully than telling the truth according to the abovementioned three criteria, are listed below.
1.4.1 Behavioral Evidence
Lying as an intentionally initiated process (1): The Activation-Decision-Construction Model (ADCM) considers lying as composed of three mental
operations (Walczyk, Roper, Seemann, & Humphrey, 2003). Confronting with the question triggers activation component in which the truth is activated and related information is currently active in working memory. In the face of activated truth, one must be attentive to whether telling that information jeopardize personal interest or not and consequently a decision to tell the truth or a lie is made. Once the decision to lie is made, the truth is inhibited by central executive and a lie is constructed. Even though activation of the truth component is generally automatic but may necessitate attentional resources to retrieve the information from long term memory depending on the accessibility of the information, it is clear that decision to and construction of a lie components are intentional.
Lying requires more cognitive resources than telling the truth (2): Examples for cognitive resources utilized in telling the truth and a lie are response latency and secondary task performance. Response latency is a good indicator of the difficulties involved in the task. It was repeatedly documented that participants start answering the questions quicker when they were required to tell the truth compared to telling a lie (Besken, 2018; Farrow et al., 2003; Vrij et al., 2008; Walczyk et al., 2003). Another way to assess the cognitive resources required to complete a task is
comparing the secondary task performance. Hu et al. (2015) showed that
performance in a secondary task suffered more while telling a lie compared to telling the truth. Thus, we can claim that lying requires more cognitive resources than telling the truth.
Lying benefits from practice (3): There are mixed results about the effect of practice on ability to lie. Some studies report that training participants to lie would not make lying get easier since lying is a cognitively demanding task and the difficulties during lying is resistant to be reduced after practice (Johnson, Barnhardt, & Zhu, 2005; Vendemia Buzan, & Green, 2005). On the other hand, some studies report that training participants to lie more frequently than telling the truth reduce their response latencies to lie (Bockstaele et al., 2012; Verschuere, Spruyt, Meijer, & Otgaar, 2011). Bockstaele et al. (2012) claims that the reason why Johnson et al. (2005) and
Vendemia et al. (2005) did not find practice effect is that their methodology did not allow participants to be optimally trained to lie such that Johnson et al. (2005) used random distribution of truth and lie trials and Vendemia et al. (2005) used 50% truth and lie trials. However, Bockstaele et al. (2012) used 50% distribution for control group and for 75% for both in truth (75% truth trials) and lie (75% trials) training groups. In the studies that used high-frequency of lie trials both reduction in the response latencies (Johnson et al., 2005) and reduction in the error percentage (Verschuere et al., 2011) were observed. It seems like when sufficient training opportunities were provided, lying benefits from practice.
1.4.2 Neurocognitive Evidence
It is possible to infer the degree of cognitive resources involved in task by analyzing the activation levels in relevant brain areas (2). Higher activation in working memory and executive control (Nunez, Casey, Egner, Hare, & Hirsch, 2005) as well as
response inhibition (bilateral ventrolateral prefrontal cortices) (Spence et al., 2001) were observed during lying compared to telling the truth (1). Interestingly, during telling a lie to others, there was higher activation both in areas indicated in social relationships (ventromedial prefrontal cortex and amygdala) and executive functions (left dorsolateral and right anterior prefrontal cortex) (Abe, Suzuki, Mori, Itoh, & Fujii, 2007).
Based on abovementioned information, we can claim that lying is an intentionally initiated process (1), requires more cognitive resources than telling the truth both cognitively and emotionally (2), and benefits from practice (3). This claim supports the premise that lying is processed more effortfully than telling the truth. Given that lying is processed more effortfully, superior memory performance for lying over telling the truth would be a reasonable expectation (Hasher & Zacks, 1979; Johnson et al., 1993).
1.4.3 Contrary Evidence
Despite these justifications, there are cases in which telling the truth produces higher memory performance than lying (Polage, 2004; Sauerland et al., 2015).
In Polage (2004), participants were asked to rate some events (e.g., hospitalized overnight, lost in a mall) from “definitely happened to me” to “definitely did not
happened to me”. Then, two weeks later, they were required to lie by changing their responses for the instructed items. For the events that they previously had rated as “definitely did not happened to me”, they needed to claim that such events had happened to them, by changing their ratings. One week later, they were given a test consisting of items that they previously changed and new items. On the last test, they were instructed to rate the events as honestly as possible. More than half of the participants changed the items they lied about before into their original ratings, which means participants remembered the real ratings more than the ratings that they lied about. In this study, the act of lying reinforced the memory for the original truthful answers.
Another conflicting piece of evidence comes from a change blindness paradigm by Sauerland et al. (2015). In this study, participants were asked to answer the questions related to their transgression history by lying to some of the predetermined questions (e.g., committed a small-scale shoplifting) and by telling the truth to the rest (e.g., stole a bike). One week later, they were interviewed with the same questionnaire they completed, but two of their truthful and deceptive answers were altered. Participants were more likely to detect the changes in their truthful answers compared to their deceptive answers. Despite the personal nature of the questions, participants were blind to the changes in their deceptive answers. A possible explanation as to why these studies manifested a rather different pattern of results will be touched upon in the discussion section.
1.5 Metamemory and Judgments of Learning
From an applied perspective, contemplating about whether the information will be remembered or not while telling deceptive information might influence behavioral outcomes. For instance, if one is not sure about remembering their lies, they might choose to tell the truth instead, whereas if one is overconfident about themselves, they might choose to tell a lie in cases where telling a lie provides a more beneficial outcome from the liar’s perspective.
The thoughts about our own memory, how confident we are that our memories are correct, how confident we are in remembering the destination or the source of a specific information later on, basically our judgments, predictions, and awareness of our memory are studied within the scope of metamemory (Alter & Oppenheimer, 2009). One of the most common methods to measure metamemory is Judgments of Learning (JOLs). These judgments are collected from participants by asking them to report on a scale of 0 to 100 (rarely from 0 to 5 or 10) how confident they are in remembering the material, their answers etc. on a memory test to be held
subsequently either item-by-item (asking their JOLs after each item) or aggregately (asking their JOLs after all items in encoding phase are completed). Then,
participants are asked to remember the items, and JOLs and memory performance are compared.
One of the methods to compare JOLs and actual memory performance is looking at the relative accuracy. For example, if the JOLs for different experimental conditions match the results of the relevant actual memory performance, we can claim that relative accuracy is high. On the other hand, if a condition with lower JOLs, produce
higher memory performance than other condition(s) and/ or if a condition with higher JOLs produce lower memory performance than other condition(s), we can claim that relative accuracy is low (Dunlosky, Mueller, & Thiede, 2016).
Generally, JOLs and actual memory performance produce results in the same direction, leading to high relative accuracy (Bjorklund & Douglas, 1997; Koriat & Goldsmith, 1996; Yue, Castel & Bjork, 2013). However, there are situations in which the pattern of JOL for a manipulation does not coincide with the actual memory performance. These cases with low relative accuracy can occur due to single dissociation between JOLs and memory. JOLs might be sensitive to the differences between conditions but actual memory performance might not differ (Frank & Kuhlman, 2016; Rhodes & Castel, 2008) or memory might be sensitive to a certain manipulation, but JOLs might not be influenced accordingly (Sungkhasettee, Friedman, and Castel (2011). Another pattern of low relative accuracy could occur due to double dissociation between JOLs and memory, i.e., metacognitive illusions. Memory performance might yield the opposite pattern of what participants predicted (Besken, 2016; 2018; Besken & Mulligan, 2013; 2014; Hirshman & Mulligan, 1991; Koriat & Ma’ayan, 2005).
One of the factors that influences JOLs is fluency (ease) of processing, subjective feeling of how easily (or difficultly) a mental operation is completed (Alter & Oppenheimer, 2009; Begg, Duft, Lalonde, Melnick, & Sanvito, 1989; Hertzog, Dunlosky, Robinson, & Kidder, 2003; Oppenheimer, 2008). Begg et al. (1989)
those words are processed more fluently. However, the actual memory performance was not always affected by fluency in the same manner: Participants indeed
remembered more fluent concrete words more than disfluent abstract words, but disfluently processed low-frequency words were remembered more than fluently processed high-frequency words.
Hertzog et al. (2003) introduced a more objective method to measure fluency of processing which is response latencies. Hertzog et al. (2003) claimed that the more fluent materials will be processed faster and therefore would have lower response latencies whereas disfluent materials would have higher response latencies due to the difficulties during encoding. In the study, participants were asked to visualize word pairs to be remembered later and pressed a key as soon as they generated a visual image. After providing JOLs, they recalled the words. Results showed that participants relied on the response latencies during providing JOLs even though response latencies were not related to recall performance. It can be deduced that fluency of processing is a good indicator of JOLs (see Rhodes, 2016; Oppenheimer, 2008 for a review), but this processing fluency may not always influence retrieval in the same manner. Instead, JOLs reflect participants’ subjective expectations of difficulty rather than actual difficulty (Oppenheimer, 2008).
1.5.1 Lying and Metamemory
Even though the terms metacognition and metamemory were used since the 1970s (Flavell, 1979; Flavell & Wellman, 1977), how lying influences metamemory is not explored (see Besken, 2018 for exception). As shown in Section 1.4.1 and 1.4.2, telling a lie operates intentionally and produces longer response latency (Walczyk et
al., 2003), influences the secondary tasks negatively (Hu et al., 2015), benefits from practice with decreased error rates (Verschuere et al., 2011) and activates more areas in the brain that are related to both emotional and cognitive processing (Abe et al., 2007), which suggests that lying requires more cognitive effort to process (Hasher & Zacks, 1979). In line with the claims about fluency of processing (Begg et al., 1989; Hertzog et al., 2003), the difference in the amount of cognitive resources involved in telling the truth and lying might influence memory and metamemory differently. That is, since lying is processed less fluently than telling the truth, the difficulties during encoding deceptive information might reduce predicted memory performance, JOLs, whereas an increase in the actual memory performance might be observed.
This hypothesis was tested by Besken (2018). In this study, 32 general information questions were used. Participants were asked to answer half of them by telling the truth and the other half by telling a plausible lie. After answering each question (all questions in Experiment 2), participants were instructed to rate their confidence in remembering their own answers on a scale of 0 to 100. Then, they performed a distractor test and a free-recall memory test in which they were asked to recall as many answers as they could in 5 minutes. Results revealed that lying is objectively more disfluent than telling the truth based on response latencies. Participants were quicker to start responding and finish typing when they were asked to tell the truth. They predicted that they would remember their truthful answer more than their deceptive answers; yet, the actual memory performance was in the opposite
performance. Difficulties while telling a lie lowered predicted memory, but the very same reasons increased memory for deceptive information.
An important factor to reinforce this evidence is that lying was studied systematically across experiments and same pattern of results were obtained. Giving participants either unlimited time (Experiment 1) or 12 seconds to complete their answers
(Experiment 2), or easing the process of lying by providing the first two letters of the answers (Experiment 3), or by providing two options to select according to the lie or truth prompts (Experiment 4) did not alter the pattern of results. Across all
experiments, although the deceptive answers were predicted to be remembered less, in reality they were remembered more than the truthful answers.
1.6 Aims of the Current Study
Since lying is a complicated phenomenon with various factors (e.g., types of lying, the content of the lies, social contexts, recurrent relationships, arousal), we restrained our focus in the current study. We aimed to investigate the underlying cognitive mechanisms between lying, memory, and metamemory rather than social aspects of lying. Thus, we asked participants to fabricate an impersonal information that is contradictory to the real information in which they would need to follow the steps in the ADCM: activation of the truth, decision to lie, construction of a plausible lie (Walczyk et al., 2003) and participants answered by typing their answers on a keyboard without any social interaction or distraction. Additionally, in the current study we focused on deliberate lying (see Section 1.1.2) by asking participants to lie and among the types of deliberate lying we focused on elaboration (See Section
1.2.1) by asking participants to create an alternative account of the reality. Generalization of the results should be addressed accordingly.
The question of whether the results obtained with semantic memory (Besken, 2018) are generalizable to episodic memory bears crucial importance, since the content of lies generally include events that the person experienced or witnessed only once. Semantic memory refers to knowledge of language which has necessarily no temporal or spatial links, whereas episodic memory refers to events that have temporal, spatial, contextual links associated with them (Tulving, 1972) and the rememberer engages in mental time travel to retrieve episodic memory (Tulving, 1985). For example, a rememberer do not have to mentally travel through time to remember “The capital of Turkey” (i.e., semantic information) whereas one must link the spatio-temporal and contextual information to remember “What they did in Ankara during their vacation” (i.e., episodic information). Considering that episodic and semantic memory share distinct mechanisms (Tulving, 1985), it is possible that lying has a different effect on episodic memory. By using novel pictures that participants would be exposed to only once, we aimed to probe participants to retrieve episodic memory traces for the material that they experienced seconds ago via mental time travel as opposed to using general knowledge questions in which participants were probed to use their prior knowledge which has no spatio-temporal referents to their prior experiences. This study aims to explore the question of whether lying has the similar effects on episodic memory to the effects documented with semantic memory. More specifically, whether memory performance of the
The second aim of this study is to explore whether the type of questions used in the testing phase, types of retrieval, alters the results. Besken (2018) utilized free-recall memory test. However, as noted above, free- and cued-recall memory tests produce different results in various paradigms. In order to understand which factors aid recalling the previously lied answers, different memory tests were provided across experiments, i.e., Experiment 1 utilized cued-recall whereas Experiment 2 utilized free-recall.
Another aim of this study is to explore how the different types of recognition tests are influenced by deception, that is how the relationship between destination and source memory would unfold with an episodic lie-generation manipulation. By its nature, lying requires two-way communication: it requires a perpetrator and a receiver. In the literature, it was documented that the ability to detect the people we received information from was higher compared to the ability to detect the people we sent information to. However, it is critical to keep a mental record of people that we either lied to or received deceptive information from, in order not to suffer from adverse consequences. Thus, lying manipulation might potentially change the direction of the relationship between destination and source memory, which has not been tested before. Additionally, participants’ predicted memory performances for destination and source conditions will be compared to their actual memory
performances. Therefore, this study aims to introduce the results of such comparisons into literature.
CHAPTER II
EXPERIMENT 1
This study aims to address whether episodic memory operates similarly to semantic memory in the case of lying. Participants were presented with pictures, and then they were asked to answer questions about these pictures. They answered half of the questions deceptively by lying and the other half honestly by telling the truth for the pictures. If episodic memory shares a similar mechanism to semantic memory in a lie-generation manipulation, we hypothesized that generating lies would produce longer response latencies, lower predicted memory performance, and higher actual memory performance compared to telling the truth.
2.1 Participants
Thirty-five students whose mother tongue is Turkish and who have a normal or corrected-to-normal vision at Bilkent University participated. The age range was 18-24 (M = 20.44, SD = 1.35, 23 females). Three of the participants were removed from
the G-Power program (Faul, Erdfelder, Buchner, & Lang, 2009). This study was approved by the Institutional Review Board of Bilkent University. Each participant was tested individually in front of a computer at Bilkent Psychology Labs. The experiment was conducted with E-Prime 2.0.
2.2 Materials
Eighty-four pictures from the internet were gathered according to the selected categories (e.g., spices, fruits, toys, flowers, musical instruments) from Van
Overschelde, Rawson, and Dunlosky (2004). The selection criteria were as follows: The category should be distinguishable among other categories. The pictures should not be emotionally or perceptually distinctive. No more than two pictures for a category were used (see Appendix A for the questions).
As the experiment aimed participants to focus on details in each picture, several people around the lab were asked to indicate whether the focused details were clear and comprehensible or not. After the pictures and their details were determined, two of the pictures were assigned for practice trials and the rest as test stimuli. The list of 82 questions were randomly distributed to two lists of 41 items to counterbalance across participants for the type of encoding (whether they will tell the truth or tell a lie about the picture).
2.3 Design
This study had a within-subjects design. Encoding condition (truth, lie) was
manipulated within-subjects, each participant answered half of the questions honestly by telling the truth and the other half deceptively by telling a lie. The question lists
were counterbalanced, across participants such that for each question, half of the participants lied whereas the other half told the truth. The order of the pictures was randomized via E-prime 2.0 for each participant. Additionally, at most, two questions from either truth or lie conditions were presented consecutively. That is, if one
question required participants to tell a lie, the following question can either require telling the truth or telling a lie, but if two questions in a row required participants to lie, the following question had to require telling the truth.
2.4 Procedure
The experiment consisted of four phases: encoding, distractor, memory test, and truth-check phases. At the encoding phase, participants were presented with pictures depicting objects and events they might encounter in daily life. They were instructed to attend to the details described below the picture since they would be asked to answer questions regarding those details. Eighty-four pictures, two practice and 82 critical trials, were presented randomly one-by-one. A question followed each picture. Half of the questions were answered by telling the truth, the other half by lying. The questions to be told the truth or lied were determined by the program and counterbalanced across subjects. Each participant answered 41 questions honestly, and the other 41 questions dishonestly. Thus, encoding condition (telling the truth and lying) was manipulated within participants.
Each picture was presented for 5 seconds, which was determined to be sufficient to read the sentence that specifies the direction of attention and search for and attend to
to see the question, the participants needed to press d for “Truth” prompts and y for “Lie” prompts by using the keyboard. This procedure was included so that
participants would pay more attention to how they would respond. After pressing the correct key, participants were presented with a question related to the detail they attended at the previous screen and were given 12 seconds to read the question and give an answer. The program recorded the time between the onset of the question and the first key-press as response latency.
The questions were always directed at the detail to which they attended while viewing the picture. So, the participants were asked to change the detail that has already been learned instead of fabricating a completely new detail. For example, at the first screen, they would be presented a picture depicting pillows on a sofa along with the sentence “Please attend to the design of the pillow on the left,” which is a map of the world (see Appendix B for exemplary pictures and answers). In the next screen, participants were asked: “What is the design printed on the pillow on the left?” If they were prompted to tell the truth, they should answer in line with what they saw in the picture (i.e., map of the world), whereas if they were prompted to tell a lie, they should answer in contrast to what they saw in the picture (e.g., any design except a map of the world). In order to control the duration of each question across participants, each question remained locked on the screen for 12 seconds, even if participants completed their responses earlier.
After 12 seconds, the program automatically proceeded onto the next screen in which “Confidence in Remembering” (“Hatırlama Eminliği” in Turkish) was written. On this screen, participants were asked to indicate on a scale of 0 to 100 how confident
they were in remembering their answer on a subsequent memory test. If they believed that they definitely would remember their answer, they were instructed to type 100 or values closer to 100, whereas if they believed that they would definitely not remember their answer, they were instructed to type 0 or values closer to 0. Participants were encouraged to use the whole scale. They were also informed about the type of questions (i.e., cued-recall) to be used in the memory test in order for them to evaluate their memory performance more accurately. Specifically, they were told that they would be given the pictures again with a mask on the detail they had attended previously and would be asked to provide the exact answer they provided before. Their predictions were recorded as Judgments of Learning, JOLs.
Metacognitive judgments are commonly tested through this measure (Begg et al., 1989; Besken, 2018; Hertzog et al., 2003; see Rhodes, 2016 for a review). After participants entered their JOLs and pressed ENTER, the next picture was seen on the screen (see Figure 2.1).
When all questions were completed, the distractor phase began. In this phase, participants were shown a matrix composed of black and white squares. The first matrix remained on the screen for 3.5 seconds, followed by a mask for 1 second. Then the second matrix, either identical or slightly different from to the first matrix, was presented. Participants were asked to indicate whether the first and second matrices were identical or not. This task was included to eliminate participants from rehearsing the answers from the first phase, as well as to keep them from answering the questions from working memory.
In the memory test phase, a cued-recall method was used. Participants were given the same pictures they saw in the encoding phase, but the critical portion of the picture that they attended to was covered with a gray mask. The pictures were presented randomly, regardless of their presentation order in the encoding phase. On the screen, only the masked pictures were presented as a cue, neither the type of encoding (i.e., truth or lie) nor the question related to the picture was provided. Participants were asked to repeat the very same answer they gave before when they saw the masked picture, which means if they told the truth, they needed to tell the truth again, and if they told a lie, they needed to repeat the exact same lie they told before (see Figure 2.2). Participants were given the opportunity to leave questions blank if they did not remember their answers. No feedback was provided upon their answers.
Figure 2.2 Schematic display of the memory test phase in Experiment 1 and truth-check phase in Experiment 1 and 2
After completing all the questions, the program proceeded onto the truth-check phase. Participants were shown the same masked pictures as provided in the memory test phase, but this time they were instructed to answer all questions truthfully
according to the unmasked picture they saw in the encoding phase. If they told the truth before and they believed that they remembered the detail in the picture
correctly, they were asked to repeat their answers again. If they told a lie, they were asked to report the correct answer instead of repeating their deceptive answers (see Figure 2.2). By doing so, we were able to investigate how telling the truth or a lie influence the memory performance for the actual information they learnt from the pictures.
2.5 Results
One question was removed from all analyses due to a problem in the experimental program (i.e., the answers were not recorded for that one question), leaving a total of
Table 1. Means (Standard Error of the Means) for Median First Key-press and Response Completion Latencies, Mean JOLs, Proportional Cued-recall and Truth-check Memory Performance for Truth and Lie Trials in Experiment 1
Truth Lie First Key-press latency (in milliseconds) 1595 (57) 2599 (99) Response Completion Latency (in milliseconds) 8835 (117) 8044 (128) JOLs 78.76 (2.94) 74.79 (3.17) Proportional Cued-recall Memory 84.73 (2.50) 76.40 (2.38) Proportional Cued-recall Memory Conditionalized upon Truth-check Memory 96.92 (0.64) 88.10 (1.57) Proportional Truth-check Memory 85.44 (2.45) 76.13 (2.11) Proportional Truth-check Memory Conditionalized upon Cued-recall Memory 97.71 (0.63) 87.85 (1.16)
