Keywords

The development of knowledge within the psychoanalytic tradition has similarities not only with research in academic psychology but also with the hermeneutic sciences. However, it also has its own distinctive character. We argue that the concepts, notions and assertions of psychoanalytic theory often constitute thought models that might be practically relevant. These models are thus theoretically anchored regulative principles that may be relevant for practice, although the aspects and relations they refer to are not always manifested. As such, they may contribute to ground psychotherapy as a practice where one strives to meet persons as openly and unprejudiced as possible. As this aim is also shared by Smedslund (2009, 2012b, 2016) in relation to his proposal of a so-called bricoleur model of clinical practice, it is pivotal to discuss the extent to which the perspectives are compatible and may join forces or not.

With respect to the bricoleur model , we acknowledge the critique of Salvatore and Valsiner (2010) and argue that Smedslund’s related efforts to advance Psycho-logic (PL) must be broadened so as to include the relevance not only of deductive reasoning but also of abductive reasoning. However, pace Smedslund (2012c) and Valsiner (2014b), we argue that these reasoning capacities may reveal not only conceptual relations but also causal relations between dispositional properties of persons. On the other hand, we also extend upon Green’s (2003, 2005) idea of psychoanalysis as consisting of at least two forms of clinical thinking. As he writes: “One of the main characteristics of clinical thought is that it is dialogical, that is, it deals not only with the patient who suffers, but also with the analyst who has the task of listening to the suffering in the couple they form …” (Green 2003, p. 29, see also Stänicke et al. 2019). The other aspect of clinical thinking requires that the traditional understanding of psychoanalytic perspectives as solely based upon experiences of clinical encounters must be broadened so as to acknowledge theoretical elaborations that may be removed from direct clinical experience (Green 2005, p. 9).

Psychology has benefitted tremendously from both of these aspects through its history. Thus, not only do we call for a revival of the importance of theoretical elaborations in line with the pioneers of modern psychology (cf. Valsiner 2012; Danziger 1990; Toulmin and Leary 1985), but psychoanalytic theory development must be acknowledged for having planted the seeds for many prospering fields of psychological research, such as developmental psychology, attachment theory, mentalization and models of short-term psychodynamic therapy. Even its often-considered rival, cognitive therapy, was developed from a psychoanalyst, such as Aaron Beck who densified psychoanalytic theory to models of relevance for short-term consultation.

Psychoanalytic Thought Models

Our proposal of characterizing psychoanalytic theory as comprising thought models mirrors our aim of demonstrating that these aspects of the theory are developed by thinking about possible dispositions of persons and that the models are thus potentially relevant for practice. Extending on recent advancements of causal dispositionalism (Mumford and Anjum 2011; Anjum and Mumford 2018a) the term “model” suits these aims as it opts for demonstrating that significant aspects of psychoanalytic theories are concerned with dispositional properties of persons whose causal manifestations may not always be manifested. However, although the properties and relations accounted for by the thought models are not necessarily manifested, they might emerge in the course of psychotherapy, and might be relevant to understand as significant parts of the psyche of unique individuals.

It should also be noted that psychoanalytic notions are models also in the sense of being metaphoric or building on analogies, and as such psychoanalytic thought models may be considered to be scientifically idealized ways of representing psychological phenomena. Of course, speaking of thought models as idealized does not mean that they concern anything ideal. On the contrary, as clinically relevant they are often concerned with properties that may dispose towards unwanted suffering. For example, we have the well-known model of psychodynamic defence mechanism that models the dispositional need of persons to secure their well-being from overwhelming anxieties in ways that might lead to more suffering.

A common response towards psychoanalytic theories, models and notions is the surprise, or sometimes even contempt, about having a theory of anything like, for instance, the Oedipus complex. This may stem from disparaging theoretical language that puts the dynamics of the human psyche up in analogy with the narratives of non-scientific imaginary literature, or it may come from being suspicious about modelling nearly all human action as if it was sexually oriented at its base.

Responding to this critique, a common reply from psychoanalysts has been that the Oedipus complex is differently understood today with the current advancements of self-psychology, relational psychoanalysis and modern object-relation theory. However, it should be noticed that psychoanalysts rarely trash their old concepts, models or theories. Rather, it is as if they are hoarding them, stacking them all up in a closet available for future use. If one complies with a Popperian research paradigm, one may argue that this is unscientific; after all, when a model is not in use anymore, and its actual applicability in the present even falsified, should not the theory be discarded? It is not that psychoanalytic concepts and theories have never been abandoned. However, traditionally, this is understood as something that ought to happen only when clinical experience, conceptual analysis and/or empirical research convincingly demonstrate the incoherency of concepts. A good example of this is the suggestion of Mahler et al. (1975) that infants normally develop through an autistic phase. That suggestion is contrary to every empirical finding in academic developmental psychology, and furthermore, it has also been difficult to demonstrate as a clinically viable interpretation. However, this example is not the typical case in psychoanalysis. More typically analysts do not discard their theoretical notions but let them rest in latency.

A typical example of psychoanalysts keeping their theories and terms is found in Christopher Bollas (1989) when he looks upon all the various psychoanalytic concepts, models and theories as living side-by-side. Accordingly, psychoanalytic theory, all the way from Freud and up to today’s advancements, can, according to Bollas (1989, p. 104), for illustrative purposes be tentatively placed within a periodical system. In this table, one finds psychoanalytic notions, concepts, models and theories like the Oedipal complex side-by-side with the models of defence mechanisms, Bion’s theory of alpha function, projective identification, bi-logic, potential space, mentalization, attachment, narcissistic needs, drive affects and death work – just to name some examples. As such, Bollas writes that psychoanalysis outnumbers any other therapeutic perspective by its repertoire of “valuable ideas”, and we concur with him. Over its history, psychoanalytic notions and models have been formed by therapists trying to put their own and their patients’ experiences as truthfully as possible into words, and by further abductive reasoning, this has been generalized into the current theories. Abductive reasoning will be explained in a bit more detail bellow when differentiating it from both deduction and induction. Importantly, this academic emphasis on conceptualizing clinical experiences as truthfully as possible supports the clinical aim of enabling oneself to recognize the same or similar phenomena if one happens to encounter them again in later phases or with another patient.

The Legacy of Sophocles and Melanie Klein

It has been 100 years ago since Freud conceptualized his feelings in a way he referred to as an Oedipal complex, and there is no reason to discard the model. Importantly, this is so even if you seldom, or never again, come across patients with the same or similar conflict. The model of the Oedipal complex is still available and may thus not only be potentially useful for future generations of clinicians but is also relevant for psychological science as part of a coherent theory of the dispositional properties of the human psyche. Thus, even in the lack of any current empirical data supporting its existence, it may still be psychologically relevant. However, we are prone to argue that the way Freud (1900, 1910) described the Oedipus complex, it became too much marked by the socio-demographic context of his upbringing. Today the Oedipus complex is more recognizable in line with Ronald Britton’s (1989) arguments that the traumatic part of being a child in a triangulated family is the experience of being left out from what the parents have together. Hence, today many analysts will find the emotional toll of the Oedipal triangular situation for the child is not so much a question of guilt and castration anxiety, but more of jealousy, envy and loneliness. Furthermore, the analyst today, in contrast to traditional psychoanalysis, will not be so much concerned with the infantile trauma of the Oedipal complex, but more on how the patient is currently managing new triangular situations, which reactivates jealousy, envy and loneliness in her contemporary life. Thus, the early constellation with parents are better thought of as a model for how these feelings are dealt with today, and accordingly, the treatment process is not so much focused on the early traumas in the past than with how to understand and deal with, and possibly accept that one may have such feelings.

Another compelling example of a psychoanalytic thought model that has been developing since several decades is the theory of projective identification. The model was first launched by Klein (1946) who characterized it not only as a defence mechanism by which children when playing with each other projected parts of themselves into others but also as the further process of identifying with this split of part of themselves. Decades later Bion (1962) described projective identification not only as a defence mechanism that could be manifested in children’s play, but also as an interpersonal dynamic, typically the analyst being like a container for split of parts of a patients personality and, hence, the analyst’s possible tendency to act out these parts towards the patients. Several decades went before Gabbard (1995) described how this interpersonal mechanism of projective identification is happening through the analyst’s vulnerability: As if the analyst carried “personal hooks” that the patient may, possible through nonverbal communication, activate in such a way that the analyst becomes the bearer of the patient’s issue. However, extending on the arguments of Lindstad (2020a) that the recent advancements of causal dispositionalism (Mumford and Anjum 2011, 2017a, b) are relevant for psychotherapy and psychotherapy research the model can be developed further. Thus, we suggest that the model of projective identification accounts for mutually manifested relational phenomena that might emerge in clinical encounters when dispositional properties of therapists and clients come together. Thus, a therapist may have properties that dispose her towards acting in ways that a patient, because of his properties, may react upon by interpreting the therapist as having properties that are actually features of himself.Footnote 1 Additionally, the therapist may be vulnerable in the sense that she is disposed to feel inclined to identify with the properties the patient interprets her as having and thus act as if these properties were originally hers.Footnote 2 Of course, if she also acts this way the attributes actually manifest as the therapist’s own, but they could not have emerged as such without meeting this particular patient. Thus, the model of projective identification accounts for the emergence of certain kinds of changes of the therapist and patient mutually manifesting from their encounter.

There are several other psychoanalytic thought models that could be discussed, but the two earlier examples will do to illustrate that psychoanalytic theories develop over time and by thinking about possible dispositions of persons. In the following, we will characterize psychoanalytic thought models a bit further and discuss its relevance in relation to Smedslund’s notion of the therapist as a bricoleur.

The Elasticity of Thought Models

Firstly, extending on the dispositionalist account of causality developed by Mumford and Anjum (2011) and Anjum and Mumford (2018a), we argue that psychoanalytic thought models are often about relations between causal dispositions of persons. This has already been demonstrated with regards to the model of projective identification but can be extended also to other models. For instance, the model of the Oedipal complex refers to potentially unmanifested dispositions of persons. For instance, a patient’s report of problems with triangulated situations may have to do with feelings of jealousy or loneliness, and together with the thought model of transference, one may consider whether the patient is also disposed of for feeling that she is neglected by the therapist. Thus, the model of the Oedipus complex may aid the therapist to formulate interpretations, however – and this is of utmost importance – such interpretations can only be tentative; if they do not facilitate the patient’s association, or are of no help, they must be discarded like any other misunderstanding.

Secondly, the models that refer to therapeutic interventions are characterized by not being constitutive, in the sense that they are not strict rules, but idealized regulative principles. For example, Killingmo (1997) has argued that the concept of neutrality does not imply that an analyst shall behave according to an ideal of being cold and inhuman. Rather, neutrality is a regulative principle that provides the therapist with an idea of an idealized encounter that she can assess her behaviour and interventions up against. If the patient asks the therapist a personal question and the therapist answers the question on face value, this may actually be what is most feasible to do when encountering some specific patient in some particular context, although it represents deviance from neutrality. Still, the concept of neutrality helps the therapist to think through why she answered the question from the patient and why it may have been correct, which may also provide a further understanding of the patient’s thoughts and feelings, and it may even be beneficial to provide this interpretation explicitly.

Thirdly, it must be stressed that psychoanalytic thought models are characterized by various kinds of elasticity. Building on earlier contentions by Sandler (1983) and Wallerstein (1984), first author (Stänicke and Stänicke 2014) has argued that psychoanalytic concepts must be elastic in the sense that the adept clinician should not be required to immediately search for a fit between the patients’ properties and the thought models but should entrust an open-minded attitude towards the patient and the psychotherapy process. Indeed, this clinical strength of elasticity must not be discarded in order to instantly meet scientific ideals. However, this collaborative chapter gives the opportunity for discussing the relevant kinds of elasticity a bit further. Although it is an apt scientific ideal for our discipline to advance as precise concepts as possible, it should also be uncontroversial that elasticity is important regarding how we speak about what the thought models are about as well as how we use the terms that name these models. If it is required to be scientifically precise in any clinical setting both therapists and patients will probably be exhausted, and the opportunity to flexibly and gradually adjust towards a common understanding will be lost.

However, there is also an important sense in which psychoanalytic thought models must be considered elastic with regards to scientific aims: As our human minds are undeniably fallible, we must be open for the possibility of having to advance our models and concepts even further in order to improve our conceptual grip on the world. This is a bit like the child’s discovery that her ideas of birds did not sufficiently differentiate them from butterflies, and is exemplified earlier (section “The Legacy of Sophocles and Melanie Klein”) both by psychoanalysts abandoning Mahler’s notion of children necessarily developing through an autistic phase and the openness for developing thought models of the Oedipus complex and projective identification further. Thus, admitting this scientifically relevant elasticity is not in conflict with striving for scientific precision. On the contrary, it encourages and facilitates curiosity. We should remind ourselves of Adorno’s (1973) argument that stubbornly setting our conceptual schemes in stone exemplifies nothing but a form of identity thinking that only reproduce already established ideologies.

Encountering Uniqueness: Statistics Don’t Get It!

In order to explain the pragmatic relevance of psycho-analytic thought models, it is also important to situate it in relation to the predominant reliance on statistical approaches in psychotherapy. This is also relevant in relation to Smedslund’s intriguing critique of this paradigm. Thus, a dominant assumption in psychotherapy research has been that clinically oriented psychological perspectives must be evaluated by testing whether they have come up with interventions that can be proven successful on an average level via statistically supported empirical experiments in the form of randomized controlled trials (RCTs). However, as argued by Smedslund (2009, 2012b, 2015, 2016), it is not only questionable whether the RCT design is equal to this task, but it is also questionable whether there is any need for setting such a task: As no person makes sense of things from the exact same position as any other, any complete set of human experiences is inevitably unique. Also, as nothing can ever become not experienced once it is experienced, all experiences are irreversibly unique, and cannot really be experimentally replicated. Moreover, as persons are continuously susceptible to change by attaching new meanings to things from within ever-evolving unique contexts, it cannot be taken for granted that persons will react in the same or similar ways on neither the same event nor similar events.

Hence, searching for context-transcending, regular and/or lawful causal relations between therapy interventions and outcome via RCTs is at best of limited value. Nonetheless, this kind of research is still predominant, and a plethora of psychotherapy perspectives have been thrown into rivalry competing for the best result on an average level, presumably because it is thought possible to circumvent the difficulties via statistical randomization procedures. By randomly assigning a high number of persons to groups subjected to different conditions, say, some are offered psychotherapy and others are not, and estimating whether the groups subsequently show statistically significant differences, one may conclude that the differences have been caused by the therapy. The viability of this conclusion depends on the two groups being similar in all other relevant respects, and it is held that this is taken care of by the randomization. However, although characteristics that are possible for people to share, say, nationality, may spread evenly in large-sized random groups, unique characteristics, say, memories, cannot. Thus, if it cannot be precluded that unique experiences are influential, and indeed, they always are, we cannot know whether the groups are as similar as required. Hence, the randomization procedure effectively throws the baby out with the bathwater. At their best RCTs can only indicate statistically whether therapists and patients have interacted in ways that happened to be relevant in the various research contexts (Cartwright and Hardie 2012; Anjum and Mumford 2018b). However, what was done and understood by the persons involved in these contexts, and how this led to this or that result, as well as how it may be done elsewhere, in the future, with the same, or with any other person, is simply left obscure. Accordingly, we comply with Smedslund’s (2009, 2016) arguments that practitioners cannot ever gain much from relying on theories tested by RCTs. However, we also revive Smedslund’s earlier argument (1991) that empirical research may provide relevant food for thought, and as such we argue that his persistent denial of the relevance of causal explanation for psychology (2004, 2012c, 2015) is off the mark.

Don’t Throw Causality Out Along with Humeanism

As discussed in more detail by the second author (Lindstad 2020a), the problem is not that psychologists have emphasized causal explanation, but that the prevailing conceptions of causality have been misleading. As such, the resurgent philosophical interest in understanding the relevance of dispositional properties for causality (Groff and Greco 2013) and the related arguments that the Humean conceptions of causality must be replaced by dispositionalism (Mumford and Anjum 2011; Anjum and Mumford 2018a) have already been offered on behalf of medicine and the health sciences in general (Anjum 2016; Kerry et al. 2012). However, it remains to clarify its implications for psychotherapy and psychotherapy research, and although this is not our main aim here, these implications are relevant for our account of psychoanalytic thought models.

According to the theory of causality presented by Hume (1739, 1748), causal relations consist of no more than that one can observe that events regularly conjoin or follow one another. Hume also presented influential ideas of a related counterfactual difference-making account on which causes are events without which their effects would not happen. On these conceptions, all we may know is that events of one kind can be observed as regularly conjoined or followed by events of another kind. Thus, the prevailing idea that RCTs is the best way to demonstrate relevant causal effects of psychotherapy (e.g. Roth and Fonagy 2005; Hollon 2006) fits the Humean conceptions like a glove as they imply that causal links must either be demonstrated by statistical evidence of correlation or by comparing the average outcome of exposure by stimuli with the average outcome of no exposure. Unfortunately, this inapt standard is not only uncritically upheld in the statement on evidence-based practice (EBPP) provided by the American Psychological Association (APA) (2006, p. 274), but probably, something very much alike the Humean conceptions is also what Smedslund (2012c, 2015) has had in mind when denying the psychological relevance of causal explanation. This would at least be reasonable in relation to his arguments about the limited clinical value of RCTs (2009, 2016), as well as his commitment (2009, 2015) to Dilthey’s (1894) valuable distinction between explaining via causal laws (Erklärung) and understanding agents’ points of view (Verstehen).

However, as argued by Lindstad (2020a), there are better ways to pave the way for an emphasis on human meaning-making than to deem causality irrelevant, not the least when there are accounts of causality that are radically opposed to Humean conceptions. Thus, though Valsiner (2014c, p. 19) is right that sticking to search for linear causality has led psychology to ignore alternative accounts of causality, it is not quite right that psychological phenomena exist in conditions where catalytic, rather than causal, processes dominate (Valsiner 2014b, p. 113, Valsiner and Brinkmann 2016, p. 83, our italics). Rather, Valsiner (2017, p. 16, our italics) is right that talk about causality must take a new form, and as such, we consider the recent advancements of causal dispositionalism (Mumford and Anjum 2011; Anjum and Mumford 2018a) a compelling substitute. In contrast to Humeanism, various accounts of dispositionalism have revived a view of causality on which causal relations rest upon the powers of dispositional properties to produce changes (Marmodoro 2010; Groff and Greco 2013). Thus, on the view developed by Mumford and Anjum (2011), rather than to think of causal relations as constituted by events that simply happen to succeed one another, causal relations are constituted by real properties that dispose towards other properties as their effects. Causal properties may thus only tend towards their effects so that these effects may never be manifested in any observable regularity (Anjum and Mumford 2018a). Thus, repeated measures of the same repeatable events are no royal road to know about causal relations. Although we may come to accept claims of causal relations on the basis of observed regularities, isolating variables is neither the only nor the best way to gain an understanding of causal relations. Rather, we need more thorough inquiries aiming to explain how and why causal effects emerge. Interestingly, these arguments are not only compatible with the abovementioned arguments of Smedslund about the limited value of RCT’s for clinical practice, but as hinted earlier, like ourselves we guess Mumford and Anjum (2018b) will join in with Valsiner’s (2014a, p. 18) outburst that “correlational data do not explain – they need explanation themselves!”.

Hence, to advance an account of psychoanalytic notions as practically relevant thought models we extend upon the arguments of Lindstad (2020a) that an apt emphasis on human meaning-making does not exclude causal explanations from Verstehen. Rather, understanding what something may mean for someone is more often than not to get to know about their causally powerful dispositional properties. For instance (in relation to Smedslund’s (2012b, 2016) frequently presented clarifications of the conditions for trust), thinking of your father as both understanding and caring when offering an old lady his seat on a crowded bus, may dispose you towards thinking that he has the disposition of being trustworthy. Although RCTs may indicate the existence of psychologically relevant causal relations, we fully agree with Smedslund (2016, p. 54) that predictions of the performance of one individual based on the average score of many individuals are less correct than predictions based on acquaintance with one individual. Even so, we agree with Smedslund that basing one’s practice on RCTs may stand in the way for engaging relevantly with unique patients here and now. What is needed instead are notions that enable clinicians to grasp relevant parts of the everchanging unique set of properties of patients in unique surroundings. However, pace Smedslund, this does not make causal explanations of what may emerge in psychotherapy irrelevant. Accordingly, the working models traditionally categorized as belonging to various psychotherapy perspectives (cognitive behavioural therapy, emotion-focused therapy, person-centred therapy, etc.) was probably not originally born out of considering any results from RCTs, but was rather based on more or less systematic reasoning on why various unique patients observed in therapy was suffering, and on how to relieve that suffering. This is also what psychoanalytic thought models are about. As such, rejecting Humean conceptions of causality does not rule out the relevance of causal explanation.

The Bricoleur and the Psycho-logician

From all appearances, psychoanalytic perspectives are clearly different from Smedslund’s. Where psychoanalytic theories offer conceptual innovations, analogies and intriguing metaphors sometimes even inspired by non-scientific imaginative literature, Smedslund have, as part of his efforts to advance Psycho-logic (PL) (Smedslund 1988, 1995, 1997, 2012c), argued that psychologists must strive for cooperationally precise conceptualizations, rigorous conceptual analysis and clarification of common sense. Relatedly, while Freud continues to inspire psychoanalytic scholars to inquire into irrational aspects of the human psyche, such as primary processes and dreams, Smedslund (1970, 2012c) has persistently argued that understandable aspects of the human mind are inherently logical. Although this implies that none of the perspectives can be reduced to the other without loss, their shared aim of grounding psychotherapy as a practice where one strives to meet persons as openly and unprejudiced as possible makes it vital to clarify their joint potential.

Towards this aim Smedslund (2009, 2012b, 2016, p. 50) has attempted to formulate a consistent position he has called the bricoleur model. In this model, the psychotherapist is regarded as a jack-of-all-trades who relies on whatever is at hand that might contribute to solving problems encountered. As Smedslund has argued that the clinical value of RCTs is limited, he has proposed three alternative resources for the bricoleur. First, like Smedslund (2016, p. 55), we regard it as uncontroversial that human beings come to know much about persons by sharing language, customs and practices, and that without sufficient knowledge in this regard psychotherapeutic processes could not get off the ground. Second, also uncontroversial, is that the aim of getting a sufficient understanding of unique others and their conditions are necessary for every psychotherapy process. However, Smedslund’s third resource is not only far more controversial but largely unheeded, that is, Psycho-logic (PL).

The two former resources are also highlighted in the definition of EBPP provided by APA (2006, p. 273), apparently to ensure that therapists consider the extent to which research evidence is relevant in the context of patient characteristics, culture and preferences. However, APA also declares that the purpose of the statement is to promote empirically supported principles and interventions (p. 273, our italics). Moreover, though APA endorses the integration of multiple types of research evidence, not only RCTs (p. 273–274), not only are all examples of approved alternatives in the statement still only empirical (quantitative and/or qualitative), but APA also upholds RCTs as the standard for drawing causal inferences about the effect of psychotherapy. Smedslund’s arguments about the limited value of RCTs, as well as his efforts to advance PL, are simply ignored. As we do not only agree with Smedslund’s abovementioned arguments about the limited practical value of RCTs but also reject the Humean misconception of RCTs like the royal road to clinically relevant causal explanation, we will discuss the relevance of this proposal of PL as a third knowledge base for the bricoleur in relation to psychoanalytic thought models.

In presenting the bricoleur model, Smedslund (2012b, c, 2016) has taken departure in his advancements of PL, characterized as an attempt to explicate a conceptual system allegedly implicit in ordinary language and common sense. He also characterizes this knowledge as something that we all know because we are human and have persistently characterized these explications as amounting to a priori knowledge. This latter aspect relates to his notorious claim that the great mass of empirical research conducted to test and validate psychological theories, perspectives and models have been pseudo-empirical , that is, the theories can be known as true without empirical testing (e.g., see Smedslund 1995). Instead, Smedslund has emphasized conceptual analysis and reflection on what we as persons take for granted about being a person in the world. Although one may not fully subscribe to this account of PL we do consent to his notion of pseudo-empirical research.

As described by Smedslund, PL seems relevant for the bricoleur for two main reasons. First, the abovementioned arguments against the emphasis on RCTs in psychotherapy research are already part of PL, and thus also the conclusion that the first and the second source of knowledge is relevant for the bricoleur. Since experiences are irreversible and persons are unique the bricoleur must get to know about the unique properties of persons and their circumstances. As demonstrated by Smedslund’s abovementioned arguments (section “Encountering Uniqueness: Statistics Don’t Get It!”) we can know this from reflecting on the properties that our psychologically relevant concepts refer to, that is, from reflecting on what it is like to be a person. There is no need for testing these assertions through empirical research. However, there is more to PL than this, which amount to the second reason PL is relevant for the bricoleur; that is, as the third source of knowledge about general characteristics of persons.

Can the Bricoleur NOT Make Use of Psychoanalytic Thought Models?

Smedslund has pushed his arguments towards two radical positions. On the one hand, he has presented PL as representing knowledge that everyone has of the general characteristics of persons (Smedslund 1995, 2016, p. 55).Footnote 3 On the other hand, Smedslund’s bricoleur model invokes an utmost wariness for overgeneralizations, and we are eager to discuss whether psychoanalytic thought models are compatible with the bricoleur model in this regard.

Given the uniqueness and importance of context in every clinical encounter, Smedslund (2016, p. 54) has argued that the bricoleur must be maximally open. Thus, citing Anderson and Goolishian’s (1992) proposal of a not-knowing position, he has argued that …

[t]he attitude of initial openness or not-knowing is necessary since there are no sufficiently reliable regularities and, therefore, all fixed advance hypotheses about a particular case will most likely be at least partly incorrect. In other words, all attempts to apply the psychologist’s way of seeing the world to the problems of the individual client will most probably be at least partly wrong. The reason is that despite similarities that might be found, the unique features of a case must also always be taken into account in looking for possible solutions. Since the unique cannot by definition be known in advance, initial openness is the only realistic approach. (Smedslund 2016, pp. 54 and 56)

As such, Smedslund (2016) has argued not only that psychological practice must go on without relying on statistically based empirical research but also that since what persons do cannot be predicted by any general theory or dealt with by any fixed techniques, a professional approach to psychological phenomena must remain a-theoretical and a-technical (2004, pp. 8–9). As such he has even confessed that if asked by other clinicians what theories he uses for dealing with the clinical encounter his honest answer is “none” (2004, p. 45)!

However, as the term “theory” is ambiguous, one may wonder what an a-theoretical stance implies. In ordinary parlance, the term often is used to refer to non-scientific, unwarranted and loose ideas, as in the expression “it was just something I came up with at the moment that might as well be wrong”. Smedslund may appear to have been suspicious also about what this rather deflated use of the term concerns when having argued that the ethical commitments of the bricoleur …

necessitates an effort to lay aside stereotypes, prejudices, theories, ready-made categories, and predictions. This comparative openness and attempt not to jump to conclusions is, by definition, necessary, in order to encounter and assimilate the uniqueness of the other individual. (2009, p. 791)

However, this is probably not what Smedslund had in mind when promoting an a-theoretical stance. Directly following the last cited sentence, he writes:

Therefore, what characterizes the competent practitioner is not the possession of vast amounts of general knowledge applicable to each new case. It is, on the contrary, an ability to discard and push into the background previous experiences and to listen to what does not fit into one’s pre-existing categories. (2009, p. 791)

Thus, the a-theoretical stance does not seem to be, at least not primarily, about the abovementioned deflated use of the term “theory”. Rather, Smedslund (2016, p. 55) seems primarily to have raised his voice against the prevailing idea that prefabricated recipes of psychotherapy interventions are generally applicable, paradigmatically exemplified by the claim of Kennair et al. (2002, p. 9) that “[though] there are variations between humans, … there also is a relatively uniform human nature [which] means that interventions that work on large groups of humans will probably work for random individuals”.

Thus, though Smedslund has argued that the bricoleur must take as little as possible for granted, and patiently retain maximal openness in talking with clients, his point is hardly to ban any act of forming ideas. Rather, the initial open attitude of not knowing is for letting new impressions organize into a coherent tentative picture in subsequent cooperative explorations of possible solutions with the client (Smedslund 2012b, p. 649; Smedslund 2016, p. 55). Thus, whenever a therapist’s focus is committed to a pre-construed theory promoting certain interpretations or procedures, whether this comes from aesthetic preferences (e.g. from preferring the metaphors psychoanalytic theory rather than the simplicity of cognitive therapy models, or opposite), from clinical experience with earlier clients, from clinical guidelines of governmental authorities or from inductive generalizations from accumulations of empirical data showing statistical differences between groups (RCTs), the therapists (Smedslund 2012b, p. 649) are concerned that this will distort the possibility of getting to know the individual person and the unique circumstances in sufficient detail. Described as such, we are not only in sympathy with Smedslund’s a-theoretical stance, but a similarly open attitude on behalf of the therapists has already been an explicit ideal for psychoanalysts for decades. At least, it seems closely related both to the notion of elasticity discussed earlier (section “The Elasticity of Thought Models”) and Bion’s (1967) notorious call to “listen without memory and desire”. However, there is also a third way of understanding Smedslund’s a-theoretical stance that we think is problematic.

As mentioned, Smedslund has not only persistently presented PL as common sense, but he has also argued that this implies that PL is a priori knowledge (Smedslund 2012b, p. 643 & 655, Smedslund 2009, p. 791), and thus that psychotherapy must involve a maximal reliance on what we all allegedly know about being human (Smedslund 2009, p. 791). Although one may (cf. Lindstad 2020b) argue that the question of whether PL is a priori knowledge must not be conflated with the question of whether PL is common sense, this issue is not in question here.Footnote 4 The relevant question here is rather why Smedslund does not regard a maximal reliance on PL as reliance on theory. Probably, he will reply that theories that are built from empirical data or in need of empirical testing are significantly different from knowledge that is valid whether or not it is empirically based (cf. Smedslund 2012b, p. 655). However, if this is true, then what are we to make of the myriad of established models, theories and perspectives of psychotherapy? Must the bricoleur reject these approaches (e.g. cognitive behavioural therapy, emotion-focused therapy, psychodynamic theory, etc.) as having come up with anything of practical value? Will psychoanalytic thought models then be of no value for the bricoleur? However, there is another option that may pop out for anyone acknowledging the notion of pseudo empirical research, highlighted by the following tripartite set of questions:

(1) Could it be that the plethora of psychotherapy perspectives are often mislabelled as theories, as any research aiming to test the assertions of these perspectives will be pseudo empirical? (2) If most assertions of most psychotherapy perspectives can be justified independently of empirical research by being deduced from the axioms, theorems and corollaries of PL, must they not be considered as valuable parts of PL? (3) If that is true, and if it is also correct that the bricoleur must rely on PL to a maximal degree, must the bricoleur not also then rely on psychoanalytic thought models to a maximal degree?

However, though we think that a direct positive answer to any of these questions is too quick, we also think they are on to something relevant. Interestingly, Smedslund seems to have hinted at similar ideas when arguing not only that in so far as treatments can be generally characterized at all, they involve features that can be derived from PL (2012c, p. 301), but also that as both the assertions of PL and the assertions of psychoanalytic theories are couched in the intentional idiom of wants, beliefs and feelings, psychoanalytic descriptions are often merely slightly more complicated formulations of the assertions of PL (2004, p. 21). Putting aside the questions of which formulations are most complicated, the psychoanalytic models, or semi-formal axioms and theorems, we believe that both perspectives might be relevant for recognizing practically relevant relations between the properties of the person and the relevant circumstances.

However, before explaining how and why we think the three abovementioned questions cannot be answered positively without further qualifications, it would be inappropriate not to mention Smedslund’s Norwegian colleague Waldemar Rognes (1996) whose significant work on PL is unfortunately not accessible for readers unfamiliar with Norwegian. However, it is relevant for our purposes for two reasons. Firstly, because Rognes’ seminal arguments that conducting empirical research to test the validity of several clinically relevant psychological models and perspectives will be no less pseudo-empirical than testing many other psychological theories. Secondly, for his seminal demonstrations of how various assertions of already established psychotherapy perspectives may be deductively derived from the axioms of (PL), e.g. Well’s (Wells 1997) cognitive model of social phobia (Rognes 2007a), as well as Killingmo’s (1999) psychoanalytic listening perspective (Rognes 1999) and Killingmo’s (1997) suggestions of a so-called rule of abstinence (Rognes 2006).

Thinking About Non-Lawful Phenomena

However, though we agree with Rognes and Smedslund that assertions of psychoanalytic theories are not always in need of empirical validation from statistically based research designs, we also think that they are not thereby necessarily common sense. Moreover, though empirical inquiry may not be needed for the justification of these theories, we do think that empirical inquiry may have a role to play for improving and enriching our thoughts, and perhaps even initiate our very first thoughts about what the theories describe. This seems to be in line with Smedslund’s claim that empirical research may provide relevant food for thought, and perhaps, needless to say, qualitative research with case studies has as such been the backbone of psychoanalytic theory. Interestingly, Smedslund (2012a, p. 668) has raised the related question: “Is it possible that some persons do not rely on all the axioms [of PL] in some situations?” Yes, we do not only think that this is possible but pace Smedslund (cf. Lindstad 2020b), we also think that it is possible that some persons (psychotherapists and patients alike) will never come to take all of PL for granted (i.e. all assertions derivable from the axioms of PL). This may be because one has simply not thought of this yet, either because it has not yet been deduced from what one already knows, or because it has not yet been needed to think of these aspects for explaining something, for instance, something one has observed in a clinical setting. This latter kind of reasoning is often called inference to the best explanation or abduction (Douven 2017).

Relatedly, what the psychoanalytic thought models are about, may never be taken for granted by anyone, although they are true descriptions of the dispositional properties of persons. As such, what these models account for may only rarely be manifested again, if ever manifested. Arguably, this is also the case for many assertions of PL. For instance, though Smedslund (2012b, p. 649) has relied on his analysis of the concept of trust for arguing that it is necessary for the bricoleur to build a trusting relationship with clients, and that part of this must involve caring for the clients, he has also been clear that this is not enough. Allegedly, the client must also experience the therapist as understanding, as having relevant know-how, as having control and as being autonomous. However, as argued by Lindstad (2020a, b), also when these further conditions are not yet in place, experiencing the therapist as caring still disposes of the clients towards trusting the therapist. Thus, that such a trusting relationship is not (yet) established, does not mean that the client’s experience of the therapist as caring is psychologically impotent, and knowing about this could be helpful for the bricoleur trying to put the other relevant conditions in place so that the relevant trust could emerge. Probably, Smedslund had something similar in mind when arguing that this is knowledge of strategy rather than of laws governing the psychological content (Smedslund 2009, p. 792). However, as noted earlier (section “Don’t Throw Causality Out Along with Humeanism”) there is a crucial difference between his understanding and ours at this point. We agree with Smedslund (2004, p. 90) that the idea of psychological laws is incompatible with an apt emphasis on Dilthey’s notion of Verstehen, and thus that the idea that everyone must act in the same way under given circumstances as if by some natural law, is flawed. However, pace Smedslund, that there are no causal laws for the bricoleur to fall back on, does not mean (cf. Lindstad 2020a) that knowledge about causal relations is psychologically irrelevant. It only means that the Humean conceptions of causality must be replaced by dispositionalism.

Interestingly, this point is anticipated by Smedslund’s frequent use of ceteris paribus clauses (“all else being equal”), e.g. “if no other wants and beliefs intervene” (Smedslund 1997, p. 25–26) and “and no other factors intervene” (p. 33–40). The point is also anticipated in several assertions suggested by Rognes’ (1996) when he spoke of tendencies in relation to his seminal work on “The Psycho-logic of Self-esteem”, e.g. “Any person has a tendency to avoid describing and talking about the negative aspects of one’s self-concept” (p. 303, our translation and italics). The two features may even be aptly combined, e.g. “If a person (P) has a negative self-concept, and no other circumstances intervene, then P will tend to perceive and think about those aspects of other’s behaviour that concern whether or not they care for P, and/or respect P, and/or understand P, and/or allow for P’s autonomy, and this tendency will increase in proportion to the negative valency of P’s self-concept” (p. 208–211, our italicsFootnote 5). These conditional claims are certainly about “Verstehen” and human meaning-making, as they concern the way P may understand and make sense of herself. However, though the use of the ceteris paribus phrases indicate that the Humean causal paradigm of “if A then always B” is inappropriate because the conditionals describe features of P and her surroundings that may be prevented, this does not mean that the conditionals defy a causal reading. Rather, that the conditionals describe something that might be prevented, shows the dispositional character of the causes involved. Thus, that the causes have a tendential nature (Mumford and Anjum 2011) is perfectly compatible with Smedslund’s (2004, p. 54) argument that the traditional understanding of causality as characterized by the paradigm of “if A then always B” is unfit for psychology.

If these arguments are up to something, they may have substantial implications for the understanding of what clinically relevant competency is. On the one hand, we agree with Smedslund (2009, 2015, 2016) that the prevailing portrayal of psychotherapy as the systematic use of psychological knowledge in such a way as it is with statistical probability leads to expected change (e.g. Kennair et al. 2014; Baker et al. 2008) is utterly misleading. Relatedly, we also agree with Rognes (1996, p. 98, our translation) that to the extent that practice may improve by increased clinical experience, this cannot be the result of precarious induction from the unavoidably limited and biased experience of any individual therapist. Indeed, one does not necessarily get wiser from having more experience. Rather, we agree with Rognes that clinical experience may provide a relevant “food for thought”, that is, it may be a resource for further relevant reflection. However, Smedslund’s and Rognes’ insistence that such reflection can only be concerned about conceptual relations and not causal ones are too limited, and furthermore, as we have argued earlier, it depends upon a flawed Humean conception of causality. Thus, practically relevant reflection may also, and perhaps more often, be concerned with clarifying possible relations between the general dispositions of persons, unique dispositional properties emerging from individual experiences, and the properties of the unique and complex contexts that persons are part of.

Moreover, though we acknowledge Smedslund’s and Rognes’ efforts to demonstrate the possibility that such assertions (theorems) may be deduced from other such assertions (axioms and/or theorems), we also acknowledge the earlier critique of Smedslund by Salvatore and Valsiner (2010, see also Valsiner 2012; Salvatore 2020), invoking the relevance of abductive reasoning. The abductive inference is often thought of as attempts to find the best explanation for some unexplained specific phenomenon either by adding and/or by modifying one’s assumptions with the aim of preserving consistency. Thus, abduction can be seen as an extension of induction in the direction of hypothesized deductions (if something is observed, but cannot be explained, thinking about a new feature could explain it). For instance, if the trust is not found, and the experience of care is in place, the lack of trust may be may explained by the lack of experiencing understanding. Or if P’s excessive concern about whether other persons are respectful or not, cannot be explained by actual disrespect, it might be because P’s negative self-concept involves the belief that he is not respectable, and that other people will also think so. Such a disposition could in other contexts perhaps also explain why P sometimes avoids other people and in other circumstances furiously expresses that he hates them etc. (Rognes 1996).

Indeed, this possibility of abductive reasoning from unique cases to general principles that could help explain the emergence of unique psychological phenomena does not only seem compatible with Rognes’ (1996) abovementioned critique of the prevailing emphasis on precarious empirical induction, but it also agrees with his proposal that clinical experience may provide practically relevant “food for thought”. As mentioned, this is also in line with the traditional understanding of psychoanalytic theories as based on psychoanalysts’ reflections on experiences from clinical work with unique patients: Psychoanalytic theory building has almost always begun with single-case studies (McLeoud 2010). However, as may have become clear from our discussions, the meta-theoretical understanding of what kind of models the psychoanalytic ones are, and thus of their scientific and pragmatic status, need further advancement. Despite the growing concern among psychoanalysts to tests their models via RCTs, quasi-experiments and other statistically supported empirical research methods, and notwithstanding that several psychoanalytic models now have been put to such test (de Maat et al. 2013; Leichsenring et al. 2015; Steinert et al. 2017), the nature of the properties that the notions and assertions of psychoanalytic theories refer to imply not only that it is more complex to test them via quantitative empirical research, but it is not needed.

There is now a lot of research that has aimed to test psychoanalytic notions and theories empirically (for an overview see Andersen et al. 1995; Westen 1999; Bornstein 2005; Solms 2018). However, such testing is complex both because the metaphoric quality of many psychoanalytic concepts are not easily converted into operational definitions needed for replicated empirical estimates, and because the terminology (vocabulary) that is used to express the concepts of the theory are meant to describe individual and unique features of persons rather than matters relevant on an average group level. Our conjecture, however, is that revealing the dispositional nature of many of the properties that the psychoanalytic concepts and thought models refer to, may also reveal that many attempts to test psychoanalytic theories empirically will be pseudo-empirical. As argued by Lindstad (2020a) this is in line with dispositionalism as it implies not only that RCTs and other kinds of correlational studies are rarely sufficient, but they are also not necessary for clarifying causal relations. Even before any causal effects have emerged, we may gain knowledge of relevant causal relations and mechanisms by reflecting systematically on the possible and impossible interplay of various dispositional properties. Statistical evidence is thus not needed if we already understand the mechanisms involved (cf. Anjum and Mumford 2018b). For example, we may know that windows might break when, say, books are thrown at them, simply by reflecting upon the possible interplay of the properties of books, throwing and windows. And if we already know about human vulnerability, there is no need to test empirically whether someone may become anxious when together with someone they experience as unpredictable. However, we also know that these are only tendencies. There is no unpreventable law to be found that windows break whenever books are thrown at them, or that all children having unpredictable parents will become anxious; say, the children may feel safe in the context of their grandparents, they may believe they are stronger or more competent than their parents, etc. Nevertheless, we may know independently of correlational studies that we should avoid being unpredictable if we want to deserve our children’s trust.

To the extent that these considerations are up to something, the prevailing attempts to generalize by statistically based inductive generalization are unwarranted, and the need for deductive and abductive alternatives is urgent. This implies that the traditional understanding of psychoanalytic perspectives as based upon experiences of clinical encounters must be broadened so as to acknowledge the relevance of theoretical elaborations that can be pushed to a level of reflection that may become removed from direct clinical experience (Green 2005, pp. 9–10). According to Green (2005), psychoanalytic theory cannot invoke experience as a raw fact to be reported naively, and thus, a mode of clinical thinking is needed that consider psychological phenomena from the angle of a specific causality that gives meaning to the movements, developments and transformations that offer themselves to psychoanalytic listening (p. 9). We are inclined to agree, but think the point can be made more precise: Psycho-analytic thought models are pragmatically relevant in the sense that they are concerned with dispositional relations that might be, but are not necessarily, manifested, and might be clarified by deductive and/or abductive reasoning.

Recognizing and Clarifying Misunderstandings

However, how does this relate to Smedslund’s so-called a-theoretical stance? How to ensure that such theoretical elaborations will not stand in the way and distort the necessary process of getting to know about the relevant unique properties of persons? As the only way to get to know a unique client is to enter the interaction with him or her with an open, maximally unprejudiced attitude (Smedslund 2012b, p. 649), how could clinical thinking ever become practically relevant? Moreover, one may wonder how much time psychotherapists should spend on getting to grips with thought models that concerns dispositional properties that may only rarely be manifested (if ever). The answer is twofold: Not too much, and not too little. Not so much that we do not get out of our armchairs to meet real people, and not so little that we never pull back to check for other possibilities than that our first impressions are correct.

Smedslund’s (2012c, p. 299) characterization of PL as a kind of “subjective unconscious” seems related to his understanding of PL as a shared common sense. He has even considered it likely that part of this knowledge reflects a shared inborn disposition to understand persons in a certain way (2012a, p. 658), as if PL was a kind of implicit and unreflective grammar that people rarely knew about, but that they could not avoid taking for granted. However, Smedslund has also wondered whether it is possible not to rely on all the axioms of PL in some situations (2012a, p. 668). We are glad he did, because, if the bricoleur’s reliance on PL is nothing more than acting by way of a human inborn habit we cannot evade, how can we know whether this purportedly shared habitus is really in touch with what is actually true about persons? The epistemological ramifications of this issue are discussed by Lindstad (2020b). Here, our focus is rather the clinically related question of whether the bricoleur relies on the axioms of PL simply because he cannot help it. However, it is one thing to demand (Smedslund 2012c, p. 300) that what PL describes should be correct about persons, it is quite another to demand that they refer to how any person automatically conceives of persons (p. 297). We have no quarrels with the former requirement, but though we (cf. Lindstad 2020b) do not deny the possibility that PL may amount to common sense, neither do we deny the possibility that persons do not always rely upon all of it. For all we know, people may never have experienced these aspects directly, nor have they ever thought of them as the best possible explanation for anything (abduction), nor as something that follows from something else they take for granted (deduction). However, though this may not always have happened, does not mean that it cannot happen.

Thus, we are reluctant about Smedslund’s description of the bricoleur as someone who automatically relies upon all PL axioms in all situations and who, simply by being a person, takes the entire system of PL (axioms and theorems) for granted as common sense. This is also why we think the tripartite set of questions presented (in section “Can the Bricoleur NOT Make Use of Psychoanalytic Thought Models?”) cannot be answered positively. What matters is not whether PL is common sense, but whether it makes sense to rely upon it, and if so, in what sense. However, in line with Smedslund’s (1988p. vii; 2012c, p. 301) characterization of PL as the result of a process of explicating, systematizing and organizing psychologically relevant information, we think Rognes (1996, pp. 406–408) was up to something when arguing that the bricoleur may gain interpretational degrees of freedom through such a process of clarifying the relations between psychologically relevant phenomena. In line with Smedslund’s (2009, p. 791) apt reminder of not jumping to conclusions in clinical settings and to patiently let new impressions organize into a coherent picture (2012b, p. 649), Rognes differentiated the bricoleur’s reliance on PL from blind, spontaneous, immediate and unreflective interpretations and interventions. Accordingly, he argued that such an explication process may provide the bricoleur with a reflective overview of various possibilities that might become actualized in concrete situations, and even sometimes of possibilities for change that must be followed for reaching certain goals. We would like to add that such a clarification process may also strengthen an apt attitude of not knowing and a related sensitivity for considering more relevant possibilities than one’s immediate first impressions. As such, the kind of explicatory process emphasized by Smedslund and Rognes may contribute to calibrate our human capacity for understanding, by strengthening our abilities to clarify misunderstandings.

To the extent that Smedslund’s and Rognes’ presentations can be reconciled in line with our synoptic discussion earlier, we are prone to argue that their views are compatible with our account of psychoanalytic thought models. However, in line with our abovementioned discussions, there are some aspects of Smedslund’s and Rognes’s accounts that we have deliberately left out in the presentation earlier, as we think they must be revised and supplemented. First, cf. Salvatore and Valsiner (2010), the process of explicating, systematizing and organizing psychologically relevant information about possible (and impossible) relations between properties of persons must not only include deductions but also abductive reasoning. Moreover, cf. Lindstad (2020a), many of these relations are not conceptual, but causal in the sense that they concern dispositions. Also, cf. Lindstad (2020b), it is not so relevant that the knowledge in question is common sense as that it makes sense.

Conclusions and Questions for Future Research

To the extent that psychoanalytic thought models can be supported either by deductive or abductive reasoning, they might provide valuable resources for the bricoleur’s proclaimed openness to use whatever is at hand. This point may be extended to most, if not all, psychotherapy perspectives, models and theories. The prevailing idea that the proper way to uncover relevant causes is to observe their regular effects, has thrown hundreds of clinically relevant perspectives, models and theories into rivalry, needlessly competing for the best results on an average level. However, as randomization procedures do not take individual experiences sufficiently into account, information about aspects that are relevant in each case is inevitably lost. Thus, in the ever-evolving unique and vastly complex contexts of psychotherapy, rather than to misconstrue psychotherapy models as competing hypotheses of regular causal relations between isolated variables, they are better characterized as compatible and/or overlapping attempts to put possible relations between individual and general dispositions of persons into words. At present, Bernardi (2003, p. 126) is still right that it is not clear to what extent the wide varieties of psychoanalytic theories coincide, are incommensurable, or contradict or complement each other. However, this seems to be no less the case for psychotherapy perspectives that are not of a psychoanalytic origin. If this is correct, a great amount of integrative theoretical work remains to be done, and for this aim, projects similar to or related to Smedslund’s seminal work on PL are highly relevant (cf. Rognes 2007b; Bergner 2004). The further upshot is that such integrative work will highlight the relevance of a capacity that is not only pivotal for psychological research, but also vital for any psychotherapy process: To take part in, and to study unique psychotherapy processes, requires that we take advantage of our capacity for critically calibrating our knowledge of possible and impossible relations between the properties of persons through thorough reflection.