What makes memory constructive a study in the serial reproduction of bartletts experiments

Article What makes memory constructive? A study in the serial reproduction of Bartlett’s experiments Culture & Psychology 2017, Vol 23(2) 186–207 ! The Author(s) 2017 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/1354067X17695759 journals.sagepub.com/home/cap Brady Wagoner Aalborg University, Denmark Abstract The claim that memory is constructive or reconstructive is no longer controversial in psychology However, in the last decades it has generally been taken to mean that our memories are inaccurate or distorted In the locus classicus of the constructive memory idea—Bartlett’s Remembering—we find a different meaning: Constructive is there understood as a future-oriented and adaptive characteristic of remembering, which can also lead to accuracy His notion of constructiveness was even earlier elaborated in relation to group dynamics in his book Psychology and Primitive Culture How did we get from one meaning of constructive to another? This question is explored through a serial reproduction analysis of experiments purporting to replicate Bartlett’s study The focus is on the transformation of terminology used to describe qualitative changes introduced by subjects into reproductions In this history a diversity of terms, coming from different intellectual sources, is gradually subsumed under the single term ‘distortion’ Thus, psychologists have reconstructed Bartlett’s work based on their own background, other influences and their own project for the discipline, illustrating the very constructive processes Bartlett theorized Keywords Constructive memory, Bartlett, experimental replication, distortion, Gestalt psychology There are two different understandings of the term ‘constructive memory’ The most dominant trend in psychology today is to take it to mean that inaccuracy and distortion are characteristic of our memories This approach has been Corresponding author: Brady Wagoner, Aalborg University, Kroghstræde 3, Aalborg Ø 9220, Denmark Email: wagoner@hum.aau.dk Wagoner 187 influenced by the metaphor of mind as a computer When compared to the literal storage of information on a hard disk the human mind was a poor performer Thus, constructive here is a fault or vice of memory In contrast, constructive has also been understood as a positive, future-oriented characteristic of remembering in that it ensures we can flexibly meet the challenges of a complex world filled with rapid change This was in fact how Bartlett (1932) conceptualized the term when he introduced it in his classic work Remembering The two meanings of constructive have been mixed in contemporary discussions of memory and at times integrated (see e.g., Schacter, 2012), though construction as distortion has taken the lead This article aims to provide some conceptual clarity to the term construction by exploring its transformations from Bartlett until the present day The article begins by describing Bartlett’s invention of the term in his studies of cultural dynamics and how it provided a frame for his famous work on remembering It proceeds to highlight the terminology he used in his experimental studies to analyze qualitative transformations found in participants’ reproductions We find a diversity of terms coming mostly from diffusionist anthropology but also Freudian psychoanalysis and the experimental psychology of his day This then sets a baseline to explore how Bartlett’s own experiments were serially reproduced by later researchers, purporting to replicate and extend his work As with Bartlett’s own analysis, the focus will be on what is retained, omitted, and changed through a series of reproductions In The Constructive Mind: Bartlett’s psychology in reconstruction (Wagoner, 2017), I provided a detailed analysis of these replications Here I concentrate my analysis on the terminology used to describe changes in reproductions and how these terms themselves changed over time This will provide both a powerful case of the very constructive processes Bartlett theorized and show how the meaning of construction itself was changed Bartlett’s constructivist approach: Between anthropology and psychology Although Bartlett always saw constructive remembering as positive, futureoriented and adaptive, he used it to describe two different but related processes This difference comes out more strongly when we compare the use of ‘constructive’ in his well-known book Remembering (1932) to the way it is employed in his earlier book Psychology and Primitive Culture (1923) The earlier work aimed to develop a general theory of cultural dynamics that later served as an analogy to develop his theory of remembering Sketching out the influences coming from anthropology will put us in a better position to understand the terms he used to analyze his experiments These experiments were first framed as ‘a contribution towards an experimental study of the process of conventionalization’ (Bartlett, 1916) and only much later as being about ‘remembering’ The term conventionalization comes from anthropology, where it described the process by which cultural elements transmitted from one cultural or social group go through a series of transformations until they arrive at a relatively stable form characteristic of the receipt group 188 Culture & Psychology 23(2) In Bartlett’s (1932) own repeated reproduction experiments, where he had participants reproduce a Native American folk story called War of the Ghosts after increasing time intervals, he found it was transformed by people living in Cambridge to look more and more like an English story Bartlett’s theory of cultural dynamics was a further development of diffusionist anthropology, especially as it was practiced by his Cambridge mentors William Haddon and WHR Rivers (see also Rosa, 1996) Haddon (1894, 1895) explored the variations and development of decorative art forms with an analogy to how Darwin had looked at the spread and evolution of species (Rolda´n, 1992) However, in contrast to Darwin, Haddon also believed there was a tendency to move from concrete figurative motifs to abstract, geometrical ones This can be seen in Figure 1, which depicts various alligator derived designs from the Chiriqui group of Central America In the top series, the alligator body is simplified to a Figure Derivations of an alligator amongst the Chiriqui (Haddon, 1895, p 171) Wagoner 189 symmetrical line and dot The line’s shape is a conventionalized representation of the alligator within the group, which can be modified in a number of different directions The middle design illustrates an elaborated version of the conventional sign for an alligator, as in D, in which there is a tail and mouth at both ends The spiral shape is a common motif found cross-culturally to provide closure to a design, such as in ionic columns (Valsiner, 2014) The bottom series shows how the design is elaborated with triangles surrounding the line representing the alligator’s body At the end of the series (F) the bodyline has disappeared and we are left with the geometric pattern of triangles This pattern is ‘reduplicated’ or ‘multiplicated,’ a process Haddon referred to as ‘the characteristic device of the decorative mind’ (quoted in Bartlett, 1932, p 182) In these designs we see simplification and elaboration, mixing and blending of forms, and the flexible adaptation of patterns to particular conditions (e.g., within a narrow band as with both series in Figure 1) These processes of diffusion are by no means unique to isolated, small communities In the article by Awad (2017), in this issue, there is an example of graffiti images developing from an event in which a veiled female protestor is beaten and striped revealing her blue bra Graffiti artists represented the event in different ways, one of which transformed the protestor into a superhero with blue bra Over time, however, a simplified image of simply a blue bra was enough to symbolize the event and solidarity to the values of the Egyptian revolution, feminism and resistance to oppression Bartlett’s own experimental studies also aptly demonstrated this process, especially using his method of serial reproduction with images Participants were shown an image from a foreign culture and then asked to reproduce it typically after 15 minutes The first participant’s reproduction was then shown to a second participant, who was instructed to the same The procedure was repeated with a third and so on, like the party game ‘telephone,’ ‘Chinese whispers’ or ‘Russian scandal’ (the game itself has taken on new names as it spreads to different groups) Figure illustrates a series of reproductions produced using this method The original stimulus was an African representation of a face with text below it reading ‘Portrait d’homme’ Like Haddon (1895), Bartlett (1932) analyzes the series as illustrating a progressive tendency towards conventional representation: There is much ‘elaboration’ of the figure through tilting the face vertically, giving it an oval and then round shape, and adding features such as eyes, eyebrows, a nose, and a mouth In the last two drawings we find ‘simplification’ taking the lead over ‘elaboration’ and we are left with a conventional figure resembling a modern day smiley face The diffusionist approach in anthropology developed as an alternative to cultural evolutionism’s idea that societies evolve through uniform stages but at different rates By contrast the diffusionists argued that a society’s complexity indicated that it had a history of contacts with other groups For example, the different kinds of burial practices found in Melanesian society showed a history of 190 Culture & Psychology 23(2) Figure Series of reproductions produced through Bartlett’s (1932) method of serial reproduction (pp 178–179) contact with groups having different customs (Rivers, 1914) Rather than emphasizing independent invention, as did cultural evolution, the diffusionists saw contact between groups as the greatest stimulus to social change and the development of new cultural forms Following this tradition Bartlett (1923) argued that an individual living in a ‘primitive’ (or what we would today call ‘traditional’) societies had little space for invention without contact with other groups: He may analyze; he may be the source of much reduplication; he may make new patterns of the old material; he may introduce peculiar interpretations; but in the actual invention of new detail he is practically helpless, unless he has access to communities outside his own and of a different culture It is this, beyond anything else, which acts as the spur to those constructive processes as a result of which new forms of social organization may be achieved; new cultures produced; and radical changes brought into being (Bartlett, 1923, p 238) Group contact often results in cultural forms being created that were not found in either group before their meeting This happens by bringing together two different and often conflicting cultural practices into a new form or what Rivers Wagoner 191 described as a ‘compromise formation,’ using psychoanalytic language In this process, the ‘mixing’ and ‘blending’ of cultural forms are widespread Bartlett (1923, p 281ff) gave the example of the growth of a new religious cult within a Native American community The process was triggered by a man who brought the drug peyote to his community after having encountered it on his travels to Oklahoma Although development of a new cult was stimulated by the introduction of a new cultural object, it grew on the foundations of the group’s old practices, such as the use of a sacred mound that was used in a buffalo dance ritual of the earlier religion Later, Christian elements like the Bible were further incorporated into the cult through other group contacts Bartlett emphasized that the growth of the cult was not planned by any single individual, but rather involved the weaving together of a number of scattered influences, a process he called ‘social constructiveness’ (Bartlett, 1923, 1928, 1932) In Psychology and Primitive Culture, Bartlett (1923) opposed the tendencies of ‘conservation’ of the past and ‘construction’ of the new While Bartlett saw ‘primitive’ or ‘traditional’ cultures to be primarily conservative, this did not mean that they held rigidly to their traditions: ‘‘it is because the group is selectively conservative that it is also plastic’’ (Bartlett, 1923, pp 151–152) Cultural patterns of the group are flexibly carried forward and adapted to new circumstances This description directly parallels Bartlett’s (1932) conceptualization of schema as organized mass of past reactions adapted to the particular demands of the individual’s present environment (Wagoner, 2013) For example, a stroke in a game of tennis is channeled through a history of previous stokes but is also flexibly adjusted to the specific context in which the new stroke must be made Rather than describing this as a ‘conservative’ tendency as he did with social groups, Bartlett (1932) begins to include it as part of a general theory of what makes remembering ‘constructive’ This was probably done to further emphasize that memory should not be thought of as a static register of the past in the form of isolated traces left on the mind/ brain, but rather as a living process that responds to environmental conditions Interestingly, while Bartlett was conducting his famous experiments on remembering he was also treating traumatized soldiers at the Cambridge hospital and reading Freud’s works for inspiration From this experience he argued that even the experimental psychology must analyze psychological response in light of their history like the clinician and ethnographer As an experimentalist, he also adapted a number of psychoanalytic terms to analyze qualitative transformations in reproductions (see below) In short, both the cultural patterns of the social group and the schema of the individual describe a way in which the past as an organized standard is adaptable to meet present demands While the first notion of ‘constructive’ highlights the flexible adaptation of a past standard to the present, the second more radical notion of construction was used by Bartlett to describe the process of welding together elements from divergent sources into a new form In relation to social groups we saw this above in Bartlett’s (1923) description of the peyote cult’s emergence, where it was given the name 192 Culture & Psychology 23(2) ‘social constructiveness’ It is here that Bartlett emphasized a group’s ‘prospect’ or project for the future instead of simply an assimilation of the new to the group’s existing cultural patterns In a similar vein, Bartlett (1958) saw original scientists and scientific research areas as ‘constructive’ in integrating different fields of research to shed new light on the phenomenon of interest Like the notion of ‘prospect’, the scientist must be able to ‘predict’ or ‘prophesize’ what lines of investigation will be fruitful in the future In regards to remembering, Bartlett emphasized the process involved much more than the rudimentary use of schema as one finds in bodily habits, such as playing tennis Instead, it is a reflective process in which the person ‘‘turn[s] around upon [one’s] own schemata and construct[s] them afresh’’ (Bartlett, 1932, p 206)—what might today be called ‘meta-cognition’ or ‘meta-memory’ Human beings not simply input and output memories; they construct them through the integration of a number of different influences in order to satisfy a present interest and elaborate a coherent meaning Bartlett (1932, pp 304–305) described the process in the following: There first occurs the arousal of an attitude, an orientation, an interest Then specific detail, either in image or in direct word form, tends to be set up Finally there is a construction of other detail in such a way as to provide a rational, or satisfactory setting for the attitude The fact that we are constantly reconstructing our memories on the basis of new relations was for Bartlett a positive characteristic in that it provided flexibility in a world full of change It allows us to make new connections between different domains of experience and thereby increases our variability of response across contexts Bartlett’s own approach is an illustrative example of constructively bringing together different streams of ideas to develop a fresh approach to the perennial question of what happens when we remember This can clearly be appreciated in the terminology he used to describe transformations of reproductions in his experiments In the list below we see a wide range of different terms, most of which come from either diffusionist anthropology or Freudian psychoanalysis There are also terms from earlier experimental research in psychology, such as ‘retention’ This is not an exhaustive list but will suffice to create a baseline to explore how replications and extensions of Bartlett’s experiments have selectively borrowed terms, changed their meaning, omitted many more, incorporated terms from other sources and invented new terms to meet their particular research contexts and questions The word ‘constructive’ can be seen as a broad term that covers all of these changes It comes from McDougall’s (1908) book on social psychology, where it describes an instinct—for example, nest building in birds Bartlett clearly gave it a much more general, metaphorical meaning: first, at the level of groups and then as an analogy to describe psychological processes, such as remembering The next section will consider which terms were selected and developed by others, and how the notion of constructive changed with them Wagoner 193 The serial reproduction of Bartlett’s experiments Bartlett’s experiments have been replicated and extended in numerous experiments spanning decades of research As such, they provide an interesting case study to explore changes in the discipline of psychology from Bartlett’s time until today In order to this, I created (what I believe to be) an exhaustive list of experimental studies, up until 1999, that claim to replicate and extend his theory using his methods of repeated or serial reproduction These studies were coded for the different terms used to describe changes in a series of reproductions of some material, each indicating some aspect or factor of memory construction Table outlines the general results of this analysis, in which Bartlett’s original study and some of the terms he used are on the top row In each row that follows the terms used in a replication study are vertically paired with Bartlett’s terms Sometimes this connection between terms was conceptual rather than indicating the use of the same word The italicized terms indicate that they were borrowed from Gestalt psychology, which makes up an important part of the analysis In what follows I will provide a more detailed and contextualized picture of what is presented in the table by describing each study and highlighting key points of conceptual transformation through the history of replications The history is grouped around three trends: (1) social group comparisons in the first decade of replications; (2) replications incorporating Gestalt psychology in the 1940s and 1950s; and (3) the gradual emergence of the psychology of memory ‘distortion’ and ‘error’ Social group comparisons Authors of the earliest studies (Maxwell, 1936; Nadel, 1937; Northway, 1936) had direct contact with Bartlett and thus it is little surprising that they were very close to his analytic approach Yet, they focused on the social and cultural aspect of his 194 Table Terminological changes in replications of Bartlett’s experiments Bartlett (1932) Retention of details Omission Simplification Elaboration Maxwell (1936) Surviving Omission Shortening Invention Addition Elaboration Northway (1936) Retention Recasting Addition Modification Nadel (1937) Descriptive details Enumeration Elaboration Rationalization Tresselt and Spragg (1941) Retention Simplification Leveling Sharpening Rationalization Taylor (1947) Retention Leveling Sharpening Omission Ward (1949) Kay (1955) Omission Correct Elaboration Distortion Conventionalization Rationalization Leveling Condensation Transposition Reversions Transference Assimilation Reduplication Conventionalisation Assimilation Regularity Sharpening Assimilation Distortion Error* Skeletonization Additions Distortion* Retention Reversing Substitution Addition Errors Omission Sharpening Gauld and Stephenson (1967) Condensation Reversals Substitution Importation (major and minor) Normalization Importation Duplication Distortion Error Haque and Sabir (1975) Leveling Sharpening Wynn and Logie (1998) Bergman and Roediger (1999) Rationalization Accurate Omission *These terms were here treated like omissions in the analysis Minor distortion Major distortion Assimilation Condensation Culture & Psychology 23(2) Johnson (1962) Simplification Simplification Omission Talland (1956) Paul (1959) Condensation Etc Importation Allport and Postman (1947) Hall (1950) Conventionalization Rationalization Importation assimilation Wagoner 195 work, asking how group membership shapes remembering This was done by comparing the reproduction chains of distinct social groups, for a story created by each researcher to potentially connect up with a range of different social interests Construction was seen in the progressive transformation of material towards group interests and norms Maxwell’s (1936) story mentioned religion, sport, society, and domestic activities and had 12 different groups serially reproduce it, including priests, physics students, soldiers, and boy-scouts His analysis first looked at how mistakes intentionally built into the story (e.g., a knife is pulled out of a person who earlier was said to be shot in the head) were ‘rationalized’ or ‘smoothed over’ by participants through omission or changing of details (rationalization was used by all the studies in the 1930s but in later studies it was omitted) Further, he compared what was common and distinct for the different social groups in relation to ‘inventions and additions,’ ‘shortening’ and the material that ‘survives’ through the series Northway (1936) compared differences in recall for children with different ages (i.e., 10 and 14 to 15 year olds) and social backgrounds (i.e., from three different Toronto schools with different traditions and socio-economic levels) A central part of her analysis focused on the ‘recasting of material’ in the serial reproduction chains of different groups ‘Recasting’ was her own terminological invention to describe when ‘‘the story is given a different form, setting, or style, or that statements are so varied, distorted or elaborated that the meaning becomes quite foreign to the original’’ (p 20) For example, in one series a death at the end of the story is selected and used to construct a story with a new setting Young children tended to recast the story much earlier than older children Moreover, with regards to the 10-year-old groups, ‘‘the children at the public school (i) recast the story earlier than the children at the private school, and (ii) were much more apt to elaborate details and to give stories showing much more diversity of form’’ (p 22) Northway further subsumed other kinds of changes discussed by Bartlett (1932) under the label ‘modifications,’ which included: ‘conventionalization,’ which is given here the specific meaning of changes towards common phraseology; ‘rationalization’ which creates a more reasonable story; ‘reversals’ where phrases are turned into their opposite; and ‘substitution’ of one reason or activity for another The last three kinds of changes were much more common with younger children Nadel (1937) was an anthropologist that aimed to bring Bartlett’s approach into his field site in Northern Nigeria He had been working among the Nupe tribe, where he was also brought into close contact with the neighboring Yoruba tribe Although the two tribes were living in the same environmental conditions with similar levels of technology, they had developed entirely different, almost antagonistic, cultures The Yoruba had an elaborate and rationalized system of deities, each with its specific functions, while Nupe religion centered on an abstract and impersonal power Yoruba art was characterized by images of a human figure, whereas the Nupe developed ornamental, decorative art This contrast parallels Haddon’s (1895) description of figurative versus abstract, geometrical designs 196 Culture & Psychology 23(2) (above) Nadel claimed the focus on integrated, concrete meanings among the Yoruba and abstract details among the Nupe manifested themselves in every aspect of each groups’ culture In relation to Bartlett’s (1932) terms a tendency towards enumerating ‘descriptive details’ becomes linked with the Nupe and meaning-oriented ‘rationalization’ with the Yoruba Nadel (1937) does not use the word ‘conventionalization’ but he sets out to study it through a comparison of the two group’s recall, finding that the Yoruba tend to strengthen rational links in the story while the Yoruba tend to ‘enumerate’ details As with Maxwell (1936) and Northway (1936), Nadel is careful to argue that while cultural patterns of the two groups clearly guide recall in a certain direction one cannot ignore the individual variability within each group Group factors in recall would continue to be of interest in the next phase, though more sporadically and less centrally focused on them Enter Gestalt psychology Replications in the 1940s and 50s showed the strong influence of Gestalt psychology While experimental psychologists had previously approached the mind as a set of mental elements that are connected through different laws (e.g., of association), the Gestalt psychologists put emphasis on a person’s total field of experience, a perspective also taken by Bartlett They stressed how a person’s spontaneous organization of the field shaped how objects appear in it, indicating a kind of construction For example, their law of Pragnanz says that mental organization tends towards regularity, symmetry, and simplicity Wulf (1922/1938) demonstrated this in a classic memory study Subjects were shown a number of geometric forms and then asked to reproduce them after increasing intervals of time (i.e., 30 seconds, 24 hours and a week) This study resembles Bartlett’s method of repeated reproduction but differs from it in that Wulf used abstract visual material without obvious social content and also provided a small reminder of the stimulus before each reproduction Wulf described changes seen in reproductions as ‘‘leveling,’’ where irregularities were smoothed out and ‘‘sharpening,’’ where some dominant feature was emphasized Figure illustrates how the zigzag feature Figure Wulf’s (1922/1938) illustration of sharpening (reproduced from Koffka, 1935, p 499) Wagoner 197 of the stimulus (the top solid line) is progressively ‘sharpened’ in repeated reproductions (i.e., in the lower dotted lines) Five participants did this Only one flattened the stimulus, because the subject saw it as a ‘broken line’ rather than a ‘zigzag’ like the other participants Wulf’s terms ‘leveling’ and ‘sharpening’ have obvious parallels to Bartlett’s notions of ‘simplification’ and ‘dominant detail’ that serve as ‘active centers’ in reconstruction, respectively Moreover, both tendencies have the effect of ‘normalizing’ the design—or what Bartlett would have called ‘conventionalizing’ The term ‘assimilation’ was used by both Bartlett and Wulf Tresselt and Spragg (1941) were the first to incorporate Gestalt terms into replications of Bartlett’s (1932) experiments They first used the method of serial reproduction (where material is passed from one person to another to reproduce it) with the folk story ‘‘The boy who tried to outwit his father’’ (from Bartlett, 1932, p 129) Second, they tested the role of specific ‘mental sets’ in remembering by priming participants with one of two texts (about pot making or religion) before giving them the material to remember on ‘religion and art’ Their analysis of both experiments employed Bartlett’s and Wulf’s terms side-by-side While the first part of their paper emphasized ‘simplification’ and ‘condensation’ (i.e., running separate lines of thought together), by the end they gave much more weight to the Gestalt notions of ‘sharpening’ and ‘leveling’ The main conclusion of their study was: ‘‘The general results obtained indicate that the reproductions of verbal prose material have much in common with the reproductions of geometric materials (i.e., as found by Wulf, 1922/1938), and are describable in terms of the same general principles’’ (p 264) Allport and Postman’s (1947) classic study of rumor transmission used Gestalt terms almost exclusively, even saying explicitly that although they employed Bartlett’s method of serial reproduction they did not feel the need to adopt his terms (Allport had in fact worked with both Bartlett in Cambridge and Gestalt psychologist Kurt Koffka in Germany as part of a two-year traveling scholarship upon finishing his PhD in 1922) In one version of their serial reproduction study, a participant was shown a picture of a subway scene in which a white man was holding a razor seems to be aggressively confronting a tall black man who seems to want to avoid conflict This participant then described the scene to a second, who told a third what they remember from it and so on Allport and Postman (1947) pointed out that accounts tended to be ‘‘leveled’’ in the sense of being shortened; ‘‘sharpened’’ where features such as the razor and the size of the black man tend to take greater prominence and are exaggerated; and ‘‘assimilated’’ to conventional prejudices, such that the razor changes from being in the hands of the white man to the black man At the final stage the material reaches aphoristic brevity such that it can be repeated by rote Ward (1949) aimed to model actual historical changes made to some material within the laboratory using the method of serial reproduction with early Macedonian coin designs He in fact found similar patterns of ‘simplification’ and ‘elaboration’ in both serial reproduction and in the historical record of Macedonian coin designs between the fourth and first centuries BC, though 198 Culture & Psychology 23(2) Bartlett himself thought this was partly a matter of luck Ward (1949) did not use Gestalt terms but explicitly made an argument against their claim that there is ‘‘a definite tendency towards regularity, symmetry and simplicity’’ (p 147) (viz the law of Pragnanz) This Gestalt prediction did not pan out in either the actual history of coin design changes or in the laboratory study The controversy over whether material autonomously changes towards good form in memory was in the end inconclusive (see Riley, 1962, for a review) The title of Hall’s (1950) article summarizes his study well: The effect of names and titles upon the serial reproduction of pictorial and verbal material He refers frequently to Bartlett, Wulf, as well as Tresselt and Spragg (1941) in his setup and discussion of results Material was presented to subjects under three conditions: one of two titles was given, or no title For example, Figure shows crosssections of serial reproduction chains of a picture under the three conditions The two title series clearly develop towards a conventional representation of the title, through a balance ‘‘struck between the demands of the title and of the general trends of simplification and regularity’’ (p 116) The theorizing of his results brought together Bartlett’s theory and Gestalt psychology: ‘‘Names and titles provide a framework for the assimilation of new experience, and they tell us something about the general structure of this background’’ (p 120) Hall (1950) also argued that, ‘‘Lack of title may lead to a gross and nonsensical distortion of the original’’ (p 11) Titles, however, could both promote or impede assimilation depending on the fit between title and material In this way, the study found general evidence for Pragnanz but also showed it ‘‘could be arrested at the conceptual boundary provided by the name’’ (p 120) Talland’s (1956) study is in continuity with the first decade of replications in taking up the question of cultural differences in serial reproduction (as the article was titled) He compared serial reproduction chains of people from France, Britain, Italy, the Netherlands, Sweden and USA for a variety of different texts, including a passage from TS Eliot on the sacrament of Catholic priesthood, a textbook description of a 15th century invasion and a report of a cricket match Like Nadel (1937), Talland found that different kinds of changes were more characteristic to certain (national) groups and particular texts For example, neutral accounts (e.g., the text about 15th century history) prompted no ‘elaboration’ from participants And ‘sharpening’ was most observed in reproductions of English speaking subjects (British or American), who were said to be dissatisfied with the original’s ‘dramatic effect’ and had a tendency to preserve effective phrases In contrast to Bartlett (1932), Talland (1956) also argued, ‘‘serial reproduction does not necessarily make for more conventional, more common place words’’ (p 80) The emergence of memory distortion By the mid-1950s the influence of Gestalt psychology was beginning to wane One still finds citations to key Gestalt works and terms such as ‘sharpening’ continue to be sporadically used However, the focus of studies shifts to testing particular Wagoner 199 Figure Serial reproductions under three conditions (Hall, 1950, p 115) aspects of Bartlett’s theory of remembering with an experimental method aimed to yield definitive yes or no answers This narrowing down of what it means to an experiment could be found in the wider discipline of psychology during this same period (Winston and Blais, 1996) Moreover, here and in the discipline at large there is an increasing reliance on inferential statistics (Danziger, 1990) In earlier decades only Taylor (1947) had centered his analysis on the use of inferential statistics, whereas it increasingly became the norm from the mid-1950s onwards In Table 1, one can also observe a reduction in the diversity of terms used to describe transformations in reproductions (with the exception of Talland (1956) which belongs much more to the earlier trends, as described above) In this period, 200 Culture & Psychology 23(2) there is a funneling of terms towards ‘distortion,’ ‘error,’ and ‘accuracy,’ which become the main concern of memory research Kay (1955) set out to further explore Bartlett’s (1932, p 83) claim that ‘‘the most general characteristic of the whole of this group of experiments (using the method of repeated reproduction) was the persistence, for any given subject, of the ‘form’ of his first reproduction.’’ He extended the method to include re-readings of the original story after each time participants had reproduced it (over six sessions) so that they could see their errors Despite the re-reading of the story, Kay found participants’ first reproduction to be amazingly persistent This study was the first to make ‘errors’ an explicit category to code reproductions, alongside ‘correct,’ ‘omissions,’ and ‘additions’ ‘Errors’ and ‘additions’ differed in that the former referred to words substituted for those in the original, while with the later there was no correspondence with the original However, Kay (1955) acknowledged that these codes ‘‘did not provide a fair impression of the accuracy of a particular version’’ (p 84) and thereby supplemented this analysis with others, including a qualitative interpretation Paul (1959) aimed to explore to what extent ‘‘the distortions and fragmentations in recall’’ (p 6) described by Bartlett (1932) could be explained by the gaps and ambiguity of the story to be remembered or its unfamiliar content He prepared an ‘explicated’ version of the Native American story War of the Ghosts as well as a text about secretaries of a similar length The easiest text to remember was the one about the secretaries followed by the explicated version of War of the Ghosts Inline with his training in personality psychology, Paul identified two general types of rememberers: ‘importers,’ who tend to add material for better integration and ‘skeletonizers,’ who tend to strip, fragment, and segregate the material The first term changes Bartlett’s (1932) notion of ‘importation’ into a personality type, while the latter is Paul’s (1959) own invention—Paul (1959) claimed these terms paralleled the neurological processes ‘recruitment’ and ‘fractionation’ identified by Hebb (1949) In a follow-up experiment, Paul (1959) had groups of either importers or skeletonizers serially reproduce a text and found they transformed it in the predicted direction Johnson’s (1962) goal was to see if material ‘qualitatively changed’ in the first reproduction would be better remembered in the second than the material that was not He thought this would show that items fitting existing schemas should be persistently retained The difference between first and second reproduction explored by coding items as ‘duplicates’ (of those in the original text), ‘omissions’ (items absent in recall), ‘sharpenings’ (where an item is given greater emphasis), and ‘normalizations’ (where an item was changed towards existing conventions) Johnson believed that if Bartlett’s theory of remembering was correct items coded as ‘sharpened’ or ‘normalized’ on the first reproduction should be better remembered on the second than items coded as ‘duplicates’ In contrast, he found ‘duplicates’ were better remembered than modified items For our purposes it is interesting to note that the Gestalt terms are used here to indicate a broader term of being ‘qualitatively changed’ This would in future studies be referred to simply as ‘distortion’ Wagoner 201 While the word ‘distortion’ does appear in a number of earlier studies it was not given central importance For example, Allport and Postman (1947) mentioned the term ‘assimilative distortion’ in relation to the role of social prejudices in remembering; Hall (1950) used it to describe what happens to a figure when a title is not given (such as in Figure 4, the far right series); Talland (1956) refers to ‘collective distortions,’ which include ‘‘cultural biases, attitudes, habits of thought, and of expression’’ (p 75) characteristic of different national groups; and although Paul (1959) characterizes Bartlett’s studies as showing distortions in recall, in his analysis he treated the code ‘distortion’ the same as ‘omissions’ None of these studies use ‘distortion’ to typify what it means to remember, either individually or socially The early appearance of the term, however, shows that it was becoming a part of the psychologist’s vocabulary It was not long before it came to define what it meant for memory to be constructive This could be clearly seen in Gauld and Stephenson’s (1967) influential replication study Gauld and Stephenson (1967) characterized Bartlett (1932) as having argued that remembering is reconstructive because his participants ‘distorted’ prose passages in recall In their replication, they aimed to show that people were deliberately inventing material to fill in gaps in their memories rather than simply trying to remember the material In other words, it was not memory as such that was reconstructive (i.e., distorting) but Bartlett’s loose task instructions The idea was to separate memory from the context in which it occurs and observe it as a mental faculty To demonstrate that ‘distortion’ or ‘error’ was not characteristic of memory, Gauld and Stephenson (1967) created a different experimental instructions emphasizing ‘accuracy’ to varying degrees Subjects reproduced the story (i.e., War of the Ghosts) immediately after having heard it and the instructions Their reproductions were then coded for ‘errors,’ which were strictly defined as extreme deviations from the original Thus, omissions, word substitutions, order changes, place name mistakes, and the many other qualitative changes explored by earlier researchers were not included Error was considered the mark of memory reconstruction Given the strict instructions, limited time interval and restricted definition of errors, it is not surprising that they found few errors They attempted to explain the errors found on the basis of a person’s ‘conscientiousness,’ as measured on a personality test Their assumptions about memory could not be more different from Bartlett (1932), who had argued that we should approach remembering as an activity that is always embedded in a social and material context, rather than a faculty isolated from other factors For this reason, Bartlett (1932) had intentionally tried to create an experimental situation closer to everyday contexts of remembering In an unpublished response to Gauld and Stephenson’s (1967) article, Bartlett (1968) commented, ‘‘I did not say, I think I did not imply that literal retrieval is impossible, but I did imply that it requires special constricting conditions’’ (p 3) He went on to point out a number of social contexts in which accuracy is given central importance such as in courts of law and exams in formal schooling In Remembering he gave the example of Swazi herdsman’s incredible memory for 202 Culture & Psychology 23(2) their cattle, which was of central social interest to them (see also Ost and Costall, 2002) In brief, for him accuracy or inaccuracy cannot be considered as general properties of memory; they depend on the social context and content of what is being remembered Rather than focusing on the general category of distortion in his analysis, Bartlett attended to a number of specific changes made to a series of reproductions By the late 1960s memory ‘distortion’ and ‘error’ had become synonyms for ‘reconstruction’ With these changing assumptions, the valuation of reconstruction also shifted from being positive (i.e., indicating flexible adaptation) to negative (i.e., indicating error) Gauld and Stephenson’s (1967) study put an end to further replication studies of Bartlett (1932) for three decades, with the exception of Haque and Sabir’s (1975) small serial reproduction study of Pakistani prejudices towards the Indian army Applying Bartlett’s own theory we can say that the image of his experiments had reached a stable form within the conventions of the newly emerged cognitive psychology In the 1990s leading cognitive psychologists, referring to Gauld and Stephenson (1967), even began to claim that Bartlett’s experiments had never been replicated (e.g., Schacter, 1996, p 320) Moreover, Bergman and Roediger (1999) followed the same assumptions as Gauld and Stephenson (1967) in framing their own replication study: ‘‘Bartlett’s (1932) famous repeated reproduction experiments, in which he found systematically increasing errors in recall from the same person tested over time, have never been successfully replicated’’ (p 937) However, they improved on Gauld and Stephenson’s (1967) methodology looking at recall at different time intervals (i.e., 15 minutes, one week and six months) and by coding for ‘accuracy,’ ‘omission,’ and both ‘major’ and ‘minor’ distortions—the latter signaling surface level changes to some unit of the story War of the Ghosts They found that progressively less of the story was remembered over time and of what was remembered a greater portion was ‘majorly distorted’ Thus, they did not find support for Gauld and Stephenson’s (1967) conclusion that Bartlett’s loose task instructions account for distortions From qualitative changes to distortion: The scientific construction of constructiveness This article has described the emergence of the idea that remembering is constructive in Bartlett’s work and detailed how this idea was subsequently transformed by successive generations of psychologists Bartlett used construction to describe the ongoing and adaptive changes introduced into responses and material by both individuals and social groups For him the fact that we could flexibly apply past experience to new contexts and integrate it with other influences increased possibilities for action in a world that is filled with constant change In other words, remembering is functional to a person’s environmental needs Moreover, he situated it between individual and group processes As Northway (1936) put it in one of the earliest replication studies, ‘‘the problem of remembering is one of Wagoner 203 ‘construction,’ belonging both to individual and social psychology, it lies on the border line where the two meet’’ (p 29) She likewise explicitly clarified that ‘construction’ is a matter of making the material meaningful within ones familiar framework, a process that cannot be reduced to either cognitive or social factors Thirty years later ‘construction’ had been disconnected from its role in meaning making and assimilated into a model of memory as a mechanically operating mental faculty, that is strictly separated from the environment Through the history of replications we see the various processes of construction that Bartlett theorized Different psychologists have selected different ‘dominant details’ upon which they reconstructed Bartlett’s approach In this process many elements are omitted and others selected and bent to fit the new configuration The earliest studies took a distinctly cultural perspective and focused almost exclusively on the role of social factors in remembering, leaving aside the personal and affective dimensions that had also been of interest to Bartlett Terms Bartlett had used to describe remembering as a general process were transformed into characteristics of certain social groups—for example, the Nupe’s focus on ‘descriptive details’ and the Yoruba’s ‘rationalization’ of the material (Nadel, 1937) (cf to Talland’s (1956) observation of English participants’ tendency to ‘sharpen’ texts in recall) By contrast, starting from personality psychology and psychoanalysis, Paul (1959) transformed Bartlett’s term ‘importation’ into a stable cognitive style or personality type Much later all these different kinds of changes are subsumed (my term) under a generic mental process of ‘distortion’ in cognitive psychology In these examples we see clearly the ‘assimilation’ of Bartlett’s experiments to the familiar social background of the researchers concerned One of the most dramatic changes to the history of replications was the ‘importation’ of Gestalt terms from the 1940s on This change nicely fits Bartlett’s description of cultural contact and more specifically ‘social constructiveness,’ whereby two streams of inquiry are welded together In this process, there is much ‘mixing’ and ‘blending’ of elements, which can be clearly seen in the variable terms researchers used from both streams Although the terms are readily adopted from Gestalt psychology, researchers used stimuli closer to Bartlett’s experiments with obvious social content Thus, we see again that adoption is a selective process, driven by interests and perceived utility Furthermore, some features were widely acknowledged (e.g the law of Pragnanz) during the 1940s and 1950s but tended to be critically discussed rather than directly accepted and employed (like the term ‘sharpening’) Although there was much borrowing of terms from Gestalt psychology as well as from other sources, we find very few inventions of new terminology As mentioned above, Bartlett (1932) himself borrowed most of his terms from diffusionist anthropology, psychoanalysis, and experimental psychology His originality involved employing them in a novel way in a new context of investigation This squarely fits his diffusionist description of invention and cultural change In the replication studies, only ‘recasting’ (Northway, 1936) and ‘skeletonization’ (Paul, 204 Culture & Psychology 23(2) 1959) appear to be truly new inventions, and the latter was in fact inspired by Hebb’s (1949) notion of ‘fractionation’ It is difficult to place the term ‘distortion’ here There does not seem to be a precise source (like Gestalt psychology) from which it is taken Instead, we see that it slowly enters the vocabulary of psychologists as a peripheral term and gradually moves to the center The growing dominance of the term is likely a result of the changing assumptions and conventions in the discipline of psychology With these changes Bartlett’s (1932) experiments on constructive remembering are ‘recast’ (to use Northway’s term) into a new form This recasting coincided with the metaphor of mind as a computer that processes information, which emerged with the rise of cognitive psychology Within this metaphor ‘meaning’ became ‘information’ and ‘construction’ became ‘processing’ (Bruner, 1990) For example, ‘schema’ is redefined as a set of ‘program instructions’ for selecting, storing, recovering, and combining information (see e.g., Minsky, 1975; Neisser, 1967) Such an approach separates the organism from its environment and leaves little room for understanding ‘reconstruction’ as more than the recombination of elements ‘out there’ according to an already existing plan (for a detailed critique see Wagoner, 2013, 2017) Moreover, it imposes a fixed standard of what memory should be (viz., literal reproduction of information items) rather than studying it according to its own logic and functions Against the standard of a computer, it is easy to show that human memory is distorted The problem with ‘distortion’ as a code is that it is too overarching a term to capture the kinds and processes of change that characterize remembering as a living human process—this is why Bartlett preferred to use a wide range of different terms (see above) (however, a similar critique could be put to his term ‘elaboration,’ which those that closely followed him felt the need to subdivide—see Table 1) Starting with a different set of assumptions, one could instead label deviations from a previous idea ‘innovation’ What makes memory constructive is the same thing that makes scientific disciplines or any other cultural group constructive: namely, the ability to incorporate and develop new ideas within an ongoing project The principle function of memory as well as scientific theories is to change our relationship to the environment in order to expand our possibilities for action In his last book, Bartlett (1958) argued that ‘‘perhaps all original ideas and developments come from the contact of subject-matter with different subject-matter, of people with different people’’ (p 147) In the history described above we see the selection and mixing of ideas from a range of sources to advance certain theoretical goals But even more than this, Bartlett stressed that original scientists must have the foresight to see what lines of research are likely to pay off To paraphrase Polanyi (1962/2012, p 143), this is both more and less than knowledge: less because it is still a guess, and more because it is foreknowledge of things unknown and at present inconceivable It is precisely this forward-looking, anticipatory character that is at the heart of human constructiveness in both individuals and social groups Wagoner 205 Acknowledgements I would like to thank Daniel L Schacter, Ignacio Bresco de Luna, and Sarah H Awad for their helpful comments on an earlier draft of this article Declaration of Conflicting Interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article Funding The author(s) received no financial support for the research, authorship, and/or publication of this article References Allport, G W., & Postman, L (1947) The psychology of rumor New York, NY: Henry Holt Awad, S H (2017) Documenting a contested memory: Symbols in the changing city space of Cairo Culture & Psychology, 23(2), 234–254 Bartlett, F C (1916) Transformations arising from repeated representation: A contribution towards an experimental study of the process of conventionalization Fellowship Dissertation Cambridge: St John’s College Bartlett, F C (1923) Psychology and primitive culture Cambridge: Cambridge University Press Bartlett, F C (1928) Social constructiveness: Pt British Journal of Psychology, 18, 388–391 Bartlett, F C (1932) Remembering: A study in experimental and social psychology Cambridge: Cambridge University Press Bartlett, F C (1958) Thinking: An experimental and social study London: George Allen & Unwin Ltd Bartlett, F C (1968) Notes on remembering Retrieved from http://www.bartlett.psychol cam.ac.uk/ Bergman, E T., & Roediger, H L (1999) Can Bartlett’s repeated reproduction experiments be replicated? Memory & Cognition, 27, 937–994 Bruner, J (1990) Acts of meaning Cambridge, MA: Harvard Danziger, K (1990) Constructing the Subject Cambridge: Cambridge University Press Gauld, A., & Stephenson, G M (1967) Some experiments related to Bartlett’s theory of remembering British Journal of Psychology, 58, 39–49 Haddon, A C (1894) The decorative art of British New Guinea: A study of papuan ethnography Dublin: The Academy House Haddon, A C (1895) Evolution in art: As illustrated by the life histories of designs London: Walter Scott Publishing Hall, K R L (1950) The effect of names and titles upon the serial reproduction of pictorial and verbal material British Journal of Psychology, 41, 109–121 Haque, A., & Sabir, M (1975) The image of the Indian army and its effects on social remembering Pakistan Journal of Psychology, 8, 55–61 Hebb, D O (1949) The organization of behavior New York, NY: Wiley 206 Culture & Psychology 23(2) Johnson, R E (1962) The retention of qualitative changes in learning Journal of Verbal Learning and Verbal Behavior, 1, 218–223 Kay, H (1955) Learning and retaining verbal material British Journal of Psychology, 46, 81–100 Koffka, K (1935) Principles of Gestalt psychology London: Routledge Maxwell, R S (1936) Remembering in different social groups British Journal of Psychology, 27, 30–40 McDougall, W (1908) An introduction to social psychology London: Methuen Minsky, M (1975) A framework for representing knowledge In P H Winston (Ed.), The psychology of computer vision (pp 99–128) New York, NY: McGraw-Hill Nadel, S F (1937) Experiments on culture psychology Africa, 10, 421–435 Neisser, U (1967) Cognitive psychology Englewood Cliffs, NJ: Prentice-Hall Northway, M L (1936) The influence of age and social group on children’s remembering British Journal of Psychology, 27, 11–29 Ost, J., & Costall, A (2002) Misremembering Bartlett: A study in serial reproduction British Journal of Psychology, 93, 243–255 Paul, I H (1959) Studies in remembering Psychological Issues, 1(2), 1–152 Polanyi, M (1962/2012) Personal knowledge London: Routledge Riley, D A (1962) Memory for form In L Postman (Ed.), Psychology in the making New York, NY: Knopf Rivers, W H R (1914) The history of Melanesian society Cambridge: Cambridge University Press Rolda´n, A A (1992) Looking at anthropology from a biological point of view: A C Haddon’s metaphors on anthropology History of the Human Sciences, 5(4), 22–32 Rosa, A (1996) Bartlett’s psycho-anthropological project Culture & Psychology, 2(4), 355–378 Schacter, D L (1996) Searching for memory: The brain, the mind and the past New York, NY: Basic books Schacter, D L (2012) Adaptive constructive processes and the future of memory American Psychologist, 67(8), 603–613 Talland, G A (1956) Cultural differences in serial reproduction The Journal of Social Psychology, 43, 75–81 Taylor, W (1947) Remembering: Some effects of language and other factors British Journal of Psychology, 38, 7–19 Tresselt, M E., & Spragg, D S (1941) Changes occurring in the serial reproduction of verbally perceived materials The Journal of Genetic Psychology, 58, 255–264 Valsiner, J (2014) An invitation to cultural psychology London: Sage Wagoner, B (2013) Bartlett’s concept of schema in reconstruction Theory & Psychology, 23(5), 553–575 Wagoner, B (2017) The constructive mind: Bartlett’s psychology in reconstruction Cambridge: CUP Ward, T H G (1949) An experiment on serial reproduction with special reference to the changes in the design of early coin types British Journal of Psychology, 39, 142–147 Winston, A S., & Blais, D L (1996) What counts as an experiment?: A transdisciplinary analysis of textbooks, 1930–1970 American Journal of Psychology, 109(4), 599616 Wulf, F (1922) Beitrauăge zur Psychologie der Gestalt; VI Uuăber die Verauănderung von Vorstellungen (Gedauăchtnis und Gestalt) Psychologische Forschung, 1, 333–373 Translated extract reprinted as ‘‘Tendencies in figural variation’’ In W D Ellis (Ed.), Wagoner 207 (1938) A source book of Gestalt psychology (pp 136–148) London: Routledge & Kegan Paul Ltd Wynn, V E., & Logic, R H (1998) The veracity of long-term memories: Did Bartlett get it right? Applied Cognitive Psychology, 12, 1–20 Author biography Brady Wagoner (PhD University of Cambridge) is Professor of Psychology and Director of the MA and PhD programs in Cultural Psychology at Aalborg University He is also co-creator of F.C Bartlett Internet Archive and associate editor of the journals Culture & Psychology and Peace & Conflict His research focuses on the cultural and constructive dimensions of the mind, in particular in relation to memory and cultural diffusion His co-edited volumes include Symbolic Transformation (Routledge, 2010), Development as a Social Process (Routledge, 2013), and Integrating Experiences (Info Age, 2015) He has most recently completed a monograph titled The Constructive Mind: Bartlett’s Psychology in Reconstruction (Cambridge, 2016) and the Handbook of Culture and Memory (Oxford, 2017)  ... cultural patterns of the social group and the schema of the individual describe a way in which the past as an organized standard is adaptable to meet present demands While the first notion of ? ?constructive? ??... was conceptual rather than indicating the use of the same word The italicized terms indicate that they were borrowed from Gestalt psychology, which makes up an important part of the analysis In. .. tilting the face vertically, giving it an oval and then round shape, and adding features such as eyes, eyebrows, a nose, and a mouth In the last two drawings we find ‘simplification’ taking the lead