=Paper=
{{Paper
|id=Vol-1688/paper-22
|storemode=property
|title=Memory Priming and User Preferences
|pdfUrl=https://ceur-ws.org/Vol-1688/paper-22.pdf
|volume=Vol-1688
|authors=Evagelia Anagnostopoulou,Efthimios Bothos,Babis Magoutas,Gregoris Mentzas
|dblpUrl=https://dblp.org/rec/conf/recsys/Anagnostopoulou16
}}
==Memory Priming and User Preferences==
https://ceur-ws.org/Vol-1688/paper-22.pdf
Memory Priming and User Preferences
Evangelia Efthimios Bothos Babis Magoutas Gregoris Mentzas
Anagnostopoulou
evagelia@mail.ntua.gr mpthim@mail.ntua.gr elbabmag@mail.ntua.gr gmentzas@mail.ntua.gr
National Technical University of Athens,
Institute of Communication and Computer Systems, Athens, Greece.
ABSTRACT [4, 5], including visual objects, goals and stereotypes. When the
In this paper we provide a preliminary analysis of the effects of targeted mental representations are activated, individual behaviour
priming on user preferences and we describe two experiments that can be altered and when properly used they can improve decision-
show such effects in test environments. Our first results making or performance on tasks [6]. With respect to
demonstrate that small stimuli which primes the process of item recommendation and personalization systems, priming is
rating leads to varying average ratings. commonly omnipresent and may lead to choices that are
intentionally or unintentionally changed. For example in [7] the
Keywords authors present an experiment where an attribute based
Recommendations; Memory effects; Priming; Human decisions. recommendation agent provided suggestions for the purchase of a
tent in a virtual e-shop. They manipulated (primed) the attributes
1. MEMORY EFFECTS (e.g. weight, durability) that were presented to participants in
Recommendation and personalization systems support human order to build the recommender model and found that participants
decisions by analysing and processing past user behaviour in tend to purchase the tents that scored higher on the presented
order to filter and highlight items that users may like the most. attributes compared to those that scored higher on the excluded
Recent approaches increasingly focus not only on past behaviour attributes. Based on the above, understanding the impact of
but also on other aspects that affect human decisions including memory effects on user preferences provides opportunities to
human cognition and the important role of cognitive biases in further improve recommender systems, which is the focus of our
information systems use and understanding [1]. The study of work. In the following section we present two experiments with
human psychology has provided a rich literature on the impact of evidences that memory effects can alter user ratings. Our
non-rational decisions that can lead to sub optimal selections. inaugural experiments show that small stimuli which primes the
Commonly, the reason is that humans make choices using process of item rating leads to varying average ratings.
heuristics that can be misleading or lack the complete information
required to reach the optimal decision. As modern information 2. EXPERIMENTS AND RESULTS
systems are increasingly characterized by richness of information We designed two small scale experiments with the purpose of
and interactive decision making, human behavioural aspects need identifying the impact of memory effects on the expression of user
to be considered and cognitive restrictions to be supported. preferences, focusing on associative and repetition priming.
Associative priming refers to stimuli that appear to be unrelated,
A cognitive element shown to affect human decision making and but due to their common appearance, one will prime the other.
potentially impact the performance of recommendation and Repetition priming concerns alternations in the content of the
personalization systems refers to “memory effects” [2]. These are short term memory, rendering information readily available. The
related to our memory system that possesses a key role in the experiments were deployed on Amazon Mechanical Turk (AMT)
decision-making process especially in cases where humans that provides access to a large user base. Past studies have shown
constantly choose among alternatives. Memories can alter our that the quality of the results compares to that of laboratory
decisions while residing in the unconscious system and can experiments when the setup is carefully explained and controlled
invoke automatic reactions although decisions seem to be rational [8]. Moreover AMT has been used in past studies of recommender
and intentional. Information related to how past decisions have systems [9]. Note that we set geographic restrictions in order to
been made is recorded in our minds and is subsequently used for involve users from the US and Europe only.
future decisions, resulting to a main distinction between memory
availability and their accessibility ([2], [3]). In order for a piece of For the purpose of the associative priming experiment we
information to be available it has to be comprehended and stored implemented a demo website where users could rate three movies
in the “long-term” memory. From this point onwards, the of varying and distinct genres {comedy, sci-fi, documentary}.
information can potentially be retrieved to support judgmental Before rating the movies, users were asked to read a short story or
decisions. However, only a small proportion of available news item which related to the genre of one of the movies: a
memories is accessible at any point in time [3], depending on the funny story (matching comedy), a story about aliens (matching
time and the context. Past studies have provided evidence that sci-fi) and a news item on drugs (matching a documentary). We
important factors determining the accessible information are the selected movies with an intuitive correlation to the stories. Our
amount of competing information learned in the same content hypothesis is that the priming effects generated when reading a
domain as well as self-generated and externally generated story or an article will have impact on user ratings. After rating
retrieval cues present at the time. the movies, users were requested to provide their preferences on
what kind of movie genres they like to watch. Based on this
A main memory effect affecting decisions is “priming” which feedback we removed from the analysis users who didn’t like the
refers to cues that activate internal mental representations while genres we selected; there were 60 remaining users (20 per group).
influencing subsequent behaviour [4]. Priming may take various Figure 1 provides an overview of the results. In the case of sci-fi
forms which have been documented in the psychology literature we detected a significant higher average rating (p<0.05) while in
Copyright held by the author(s).
the case of the documentary there was some evidence of 4. CONCLUSIONS
significance (p<0.1)1. In our preliminary study we formulated the problem of
understanding memory priming effects in recommendation and
personalization systems. Our first results from two experiments
support the hypothesis that users are affected by priming stimuli.
Our next step is to better understand such effects and examine
how we can leverage them in real life systems. We are interested
in supporting sustainable living decisions through persuasive
recommendation systems. More specifically we focus on nudging
users towards green transportation choices and centre on two main
questions: i) how can we leverage environmental goal priming
with recommendations of personalized pro-environmental
Figure 1: Average ratings according to the story read. messages in travel planning applications? ii) how can we
understand and leverage the primes users receive in a given
For the purpose of the repetition priming experiment we followed
context to suggest routes that are environmentally friendly and
a similar approach with users rating a set of three movies of
increase their chances of being selected?
varying and distinct genres {comedy, action, thriller} at a demo
website. Users were divided in two groups: those who rated This research has been partially funded by the EC project
movies and afterwards provided feedback on the movies genres OPTIMUM (H2020 grant agreement no. 636160-2).
they like and those who first provided feedback on the movie
genres they like and then rated the movies. We removed from the 5. REFERENCES
analysis users who didn’t like the genres we selected; there were [1] Fleischmann, M., Amirpur, M., Benlian, A., & Hess, T. 2014.
80 remaining users (40 per group). Our hypothesis is that the Cognitive biases in information systems research: a scientometric
priming caused when users were thinking over their preferences analysis. ECIS, AIS.
yields differences in the provided ratings. Figure 2 presents an [2] Del Missier, F. 2014. Memory and Decision Making: From Basic
overview of the results. We identified a trend for lower ratings Cognitive Research to Design Issues. DMRS (8-13).
when users provided their preferences after rating the movies, [3] Endel T. Pearlstone, Z. 1966. Availability versus accessibility of
information in memory for words. Verbal Learning and Verbal
although the differences were not significant. Behavior 5 4, 381-391.
[4] Bargh J.A. and Chartrand T.L. 2000. Studying the mind in
themiddle: A practical guide to priming and automaticity research.
Cambridge University: New York.
[5] Bargh J.A., Chen M., Burrows L. 1996. Automaticity of social
behavior: Direct effects of trait construct and stereotype activation
on action. Personality and Social Psychology, 71 2, 230-244.
[6] Galinsky A.D. and Moskowitz G.B. 2000. Counterfactuals as
Behavioral Primes: Priming the Simulation Heuristic and
Consideration of Alternatives. Journal of Experimental Social
Psychology, 36 4, 384-409.
[7] Häubl, G, and Murray K. 2001. Recommending or persuading?: the
Figure 2: Average ratings for the two experimental groups impact of a shopping agent's algorithm on user behavior." Electronic
(stating genre preferences before or after rating movies). Commerce. ACM, 2001.
3. RELATED WORK [8] Paolacci, G., Chandler, J., Ipeirotis, P. G. 2010. Running
experiments on amazon mechanical turk. Judgment and Decision
Recent approaches that study cognitive biases and memory effects Making 5, 411–419.
include the following. Dennis et al. [10], study the effects of [9] Bostandjiev, S., O’Donovan, J., Hollerer, T. 2012. Tasteweights: a
“achievement priming” on user creativity and find evidence of visual interactive hybrid recommender system. RecSys, 35–42.
increased number of unique ideas in virtual brainstorming teams. ACM.
Hsieh et al. [11] build a recommendation model based on the [10] Dennis, A. R., Minas, R. K., & Bhagwatwar, A. P. 2013. Sparking
interest development theory. They identify user interests as these creativity: improving electronic brainstorming with individual
are expressed in social media and subsequently improve news cognitive priming. Journal of Management Information Systems, 29,
recommendations. Cosley et al. [12] and Adomavicius et al. [13] 4, 195-216.
study anchoring effects on human decisions to detect cognitive [11] Hsieh, C. K., Yang, L., Wei, H., Naaman, M., Estrin, D. 2016.
biases that skew the preferences users express when rating items. Immersive Recommendation: News and Event Recommendations
Using Personal Digital Traces. In WWW 51-62. ACM.
Schnabel et al. [14] leverage the short-memory effect and propose
[12] Cosley, D., Albert, S., Konstan, J., Riedl, J. 2003. Is seeing
a recommendation interface that makes use of shortlists and believing?: how recommender system interfaces affect users'
increases user satisfaction in a movie recommendation task. opinions. In CHI, 585-592. ACM.
Pinder et al. [15] suggest the use of subliminal priming for [13] Adomavicius, G., Bockstedt, J. C., Curley, S. P., & Zhang, J. 2013.
behavioral changes through priming of the unconscious system. Do recommender systems manipulate consumer preferences? A
The work described in our paper focuses on memory effects study of anchoring effects. Information Systems Research, 24, 4,
related to memory priming. 956-975.
[14] Schnabel, T., Bennett, P. N., Dumais, S. T., Joachims, T. 2016.
Using shortlists to support decision making and improve
recommender system performance. In WWW pp. 987-997. ACM.
1 Based on pairwise t-tests as performed in previous studies, see [15] Pinder, C., Vermeulen, J., Beale, R., & Hendley, R. 2015. Exploring
Nonconscious Behaviour Change Interventions on Mobile Devices.
e.g. [13]. In CHI, 1010-1017. ACM.