facebook
  • Symposium at The 24th ESCOP Conference, Sheffield, United Kingdom, September 1 - 6, 2025
  •  

    Symposium at The 24th ESCOP Conference (September 1 - 6, 2025): Can We Retrieve Information Based on Relational Similarity? 

     

    Theoretical and computational models of human memory predict that situations sharing common perceptual features are more likely to be retrieved. Empirical research on analogy, however, provides conflicting data on the roles of perceptual (surface) and relational (structure) similarity. These contradictory outcomes suggest that further investigation on the roles of surface and relational similarity in memory is needed. Bridging the two fields, in this symposium, we will present findings that not only is episodic relational retrieval possible but under some conditions, it can be preferred. We will discuss the role that structural similarity and abstract schemas play in the retrieval of autobiographic memories while surface and structural similarity are not correlated. The ADAPTER (As Deep As Possible Target Encoding and Retrieval) model will be presented as it can be used to explain the contradictory findings regarding
    the roles of surface and structural similarity in retrieval. Finally, some methodological alternatives for capturing relational retrieval of complex material will be considered. The closing discussion will focus on the role of relational similarity in memory retrieval and propose further directions in the field.

     

    An attempt to study the gap in relational retrieval differently 

     

    Penka Hristova, Patric Pförtner 

     

    New Bulgarian University, Department of Cognitive Science and Psychology, 21 Montevideo Street, Sofia, Bulgaria

     

    Email: phristova@cogs.nbu.bg 

                f85787@students.nbu.bg 

     

    The retrieval gap research employs several experimental procedures to show that humans fail to retrieve relationally similar information when object similarity is missing. The two main paradigms require recall and analogy generation and yield mixed results. The contradictions are usually explained by the inability of these paradigms to discriminate fabricated ("fake") analogies from episodic or autobiographical recall. This presents a methodological pitfall as memory is a constructive and contextually sensitive process and therefore different responses may be expected under different occasions, particularly for recall of relatively complex situations that had happened in the personal past. Furthermore, retrieval of relations is achieved with less complex stimuli, such as word-pair lists, in better-controlled recall and recognition memory paradigms, especially when cued by typical exemplars of given relations.Experiment 1 investigates whether event segmentation can be an alternative way to capture the effect of relational similarity. Experiment 2 focuses on the role of expertise in generating analogies in chess while controlling for the source of analogy (past or fantasies) and the perception of similarity, such as pixel overlap.

     

    Retrieval of relationally similar events

     

    Margarita Pavlova

     

    New Bulgarian University, Department of Cognitive Science and Psychology, 21 Montevideo Street, Sofia, Bulgaria

     

    Email: mvpavlova@nbu.bg 

     

    Classical computational models of memory predict that events that share common surface features are more likely to be retrieved (McClelland, 1995; Metcalfe, 1990; Murdock, 1995). Similarly, research on analogical reasoning demonstrates that while people may judge relational similarity as more sound for making inference, they tend to recall surface-similar information (Gentner et al., 1993; Trench & Minervino, 2015). On the other hand, the AMBR model of analogy-making and memory (Kokinov, 1988; Kokinov & Petrov, 2001) predicts that analogical episodes are more likely to be recalled than superficially similar ones. This prediction stems from the fact that AMBR represents the relations among episodes' elements, and not lists of features with no internal structure, and views recall as a reconstruction by analogy. Two sets of experiments show that analogical episodes are more probable sources of intrusions than surface similar or dissimilar ones using verbal (Feldman & Kokinov, 2009) and non-verbal materials (Pavlova & Kokinov, 2014). The results of these studies are discussed in a broader view of the role of relational similarity and the representation of structure in memory retrieval.  

     

    The ADAPTER model: As Deep As Possible Target Encoding and Retrieval

     

    Emmanuel Sander

     

    Université de Genève, Faculté de Psychologie et Sciences de l'Education, Equipe IDEA, 1211 Genève, Switzerland

     

    Email: emmanuel.sander@unige.ch 

     

    Making an analogy typically requires retrieving a source analog from memory. However, the literature presents seemingly contradictory findings regarding the roles of surface and structural similarities in retrieval. In this presentation, we introduce a conceptual model of analogical retrieval (ADAPTER: As Deep As Possible Target Encoding and Retrieval), which provides a unified account of both successes and failures in driving structurally-based retrievals. According to the model, encoding relies on the deepest available categories (with superordinate relational categories > familiar relational categories > entity categories) and constrains retrieval. The success or failure of retrieving distant analogs is primarily explained by the availability of familiar relational categories (e.g., making up an excuse) that participants can use to encode the structure of the situations. Other factors (e.g., encoding contexts) integrated into the model account for cases in which familiar relational categories fail to support structurally-based retrieval. We illustrate how the model can be used to explain previous findings post hoc and generate novel predictions regarding patterns of structurally-based retrievals.

     

    Relational categories bridge the retrieval gap: an empirical study

     

    Lucas Raynal

     

    Université de Genève, Faculté de Psychologie et Sciences de l'Education, Equipe IDEA, 1211 Genève, Switzerland

     

    Email : lucas.raynal@unige.ch 

     

    I will present an empirical study showing that non-experts are not doomed to relying on surface similarities during analogical retrieval. Participants were presented with anecdotes embodying familiar relational categories (e.g., reacting to a missed gift) and asked to report potential retrievals of personal memories. In the first experiment, the target cues preserved surface features shared by many instances of the category (e.g., reacting to a missed gift involving a cloth), causing many close analogs to compete for retrieval alongside distant analogs. In Experiment 2, a more direct competition was created between surface matches and distant analogs by using target cues involving uncommon associations between surface features and structure (e.g., reacting to a missed gift involving a school supply). The findings showed that close analogs are preferentially retrieved over distant analogs when both are available (Experiment 1), but distant analogs overcome surface matches when close analogs are sparser (Experiment 2). These results suggest that structural similarities are the primary determinant of retrieval mediated by relational categories, with surface similarities playing a complementary role in selecting a specific instance.

     

    Discussant/Moderator  

    Vencislav Popov, University of Zurich, Binzmühlestrasse 14, Box 5, CH-8050 Zürich

     

     

    Learn more about the conference here: https://www.escop.eu/events/conference/24th-escop-conference

     

     

  • Analogy 2024: The Fifth International Conference on Analogy July 23, 2024
  •  

    ANALOGY 2024: THE FIFTH INTERNATIONAL CONFERENCE ON ANALOGY

    July 23, 2024

     

    The Analogy 2024, the Fifth International Conference on Analogy took place at the University of Amsterdam on July 23, 2024. The first and second analogy conferences were held in Sofia, Bulgaria in 1998 and 2009, the third in Dijon, France in 2013, and the fourth in Paris, France in 2017.

     

    The aim of this interdisciplinary conference was to bring together the analogy researchers in person and to re-establish regularity in the conference meetings. A scientific goal of the conference was to stimulate collaborations among the analogy researchers, to inspire novel collaborations and strengthen existing relationships. The conference aimed to advance understanding of the role of analogy in learning and education, the place of structured-based cognition in other cognitive processes, and the implications of research on LLMs on understanding the cognitive processes of analogy-making. The conference concluded with a closing discussion aimed to propose an overview of where analogy research currently stands and potential future directions for both analogy research and the research community. Sixty-five people participated in the conference – both from the analogy research community and the broader cognitive science community. 

     

    The program included 3 invited symposia, 7 flash-talks, and 26 poster presentations by analogy researcher across the Globe.

     

    Invited symposia:

    • Has Analogical Reasoning Emerged in LLMs?
    • Computational Models of Structure-based Cognition
    • Analogy in Learning and Education

     

    The conference was co-organized by Claire Stevenson (University of Amsterdam) and Margarita Pavlova (New Bulgarian University)

     

    See the full program and read the abstracts here. https://sites.google.com/site/analogylist/analogy-2024/book-of-abstracts?authuser=0

     

    Learn more about the conference here. https://sites.google.com/site/analogylist/analogy-2024?authuser=0

     

     

  • 10th International Congress on Interpersonal Acceptance-Rejection June 26 - 29, 2024
  • 10-TH INTERNATIONAL CONGRESS ON INTERPERSONAL ACCEPTANCE-REJECTION

    June 26 - 29, 2024

     

    The 10th International Congress on Interpersonal Acceptance-Rejection, will be held from the 26th through 29th of June 2024 at the New Bulgarian University in Sofia, Bulgaria. This prestigious event that brings together scholars, researchers, and practitioners from all over the world is organised by the International Society for Interpersonal Acceptance-Rejection (ISIPAR) and The Department of Cognitive Science and Psychology at New Bulgarian University.

     

    The congress is a stimulating platform for exchanging cutting-edge research, engaging in thought-provoking discussions, and fostering collaborations. It is an opportunity to learn from renowned experts, share innovative ideas, and engage in intellectual discourse in shaping the future of the field of interpersonal acceptance and rejection. Participating in this event is a great opportunity to meet, exchange ideas and knowledge, collaborate toward developing new perspectives and solutions, as well as to network and build relationships with other like-minded professionals.

     

    We are confident that the warm and welcoming environment of Sofia, a city renowned for its rich cultural heritage and hospitable atmosphere, will create an unforgettable experience for all of you. Apart from the enriching academic program, the congress also offers ample opportunities for networking, social interactions, and cultural exchanges, providing a unique platform for building connections, friendships, and collaborations that can last a lifetime. 

     

    Learn more about the conference here: https://www.iciar2024.com/