As I began to write these lines, I was struck by how haphazard the factors were that shaped my intellectual interests, and later, how unpredictable the influences are that make my research field develop in unforeseeable directions.

My parental home gave no clue to my future academic career. My mother came from Vienna and valued education to my benefit. My father was a very practical, self-made mountain man, deeply suspicious of any university degree other than medicine or law. He had clear examples to bolster his bias. The most glaring example was the son of Adolf Lorenz, his best man at my parents’ wedding in Vienna. He was a highly respected doctor and a regular guest with his family in my father’s mountain lodge. His eldest son had a proper medical degree. The younger son had also studied, but patently something overly useless, to put it mildly. Young Konrad had nothing better to do then spend time with a flock of grey geese and take them daily for a morning swim in the nearby lake. That’s what a degree in zoology does to you.

On my mother’s insistence, I received higher education in a boarding school, a day’s journey from home. In the final year we had one hour of philosophy and psychology a week, lodged between Kant and Freud. Rumor had it that Wittgenstein was really interesting. So, I teamed up with a whiz kid from the year below and we worked our way through Wittgenstein’s Tractatus, with the collateral benefit of some familiarity with set theory. It also left me with an attraction to formal logic. I suspect this stemmed from a desire to save me having to argue with people and just point to the logical necessity of my claims. Doesn’t work that way, I later learned. Looking back, I realized that my choice of psychology gave me a similar benefit; I can hide behind data rather than face pure argument.

After reading Wittgenstein my vague intention was to study philosophy. My intention was thwarted by a guest lecture of a famous French philosopher at the local university. Not by the speaker, but by the local professor and his long-winded and dull introduction. I was done with philosophy at this university! So I considered psychology, as we were told it was the closest to philosophy. In the end I signed up for psychology and for mathematics as a minor. But why suddenly maths?

Between high school and university, I went through my year’s army service. It was timed to end a month before registration at university. But it was the year after the Russian invasion of Czechoslovakia (in 1968). A day before our discharge, the CIA (it was rumored) had informed Austria of Russia’s intention to take a shortcut through Austria to straighten out Tito in Yugoslavia. To ease the transgression their plan was to knock out the Austrian command center in Vienna in a swift commando action. The new recruits had no training. So we were retained and for the night placed around headquarters armed with a machine gun and orders to shoot anyone who came running. In the morning the emergency was over, but we were to serve for another month, just in case.

Out of the army, but late; registration had started at universities. I had made up my mind to take psychology but learned at registration that a minor subject had to be chosen too … absolute blank. So the registration officer helped me through the list of possible subjects with nothing really standing out. At the end: philosophy (yeah, but), maths (maybe), geography (certainly not). Then my short-term memory still had the penultimate item “mathematics” on offer. That’s how I ended up with that as my minor. Further surprise: some introductory seminars in philosophy had to be taken. After all, I was to be awarded a “Doctor of Philosophy.” No BA or MA but straight to the doctorate in our medieval system. Fortunately, philosophy at Salzburg had been enriched since my high school visit. A second professor now offered Logic and Philosophy of Science and I ended up taking more philosophy than required.

Psychology at Salzburg was a small department. Salzburg University, founded in 1622, had come close to extinction over the Napoleonic wars and was re-established in 1962, widening access to higher education in the course of the economic upturn across Western Europe. One full professor, Josef Revers, called himself a phenomenological psychologist and offered seminars on Dostoyevsky’s The Brothers Karamazov. He was open-minded about psychoanalysis and could pride himself of having introduced Freudian psychoanalysis to Austrian universities. The year after, a chair of experimental social psychology was established, and finally one for psychoanalysis after I had departed.

It’s worth noting that this was the period of prominence of the Frankfurt School (Adorno & Horkheimer). We had lively discussions about the politics and scientific foundation of psychoanalysis, which strengthened my resolve to stay with experimental psychology. Thinking ahead for a potential thesis topic, I decided on concept formation. But the experimental work looked dreary: a wug is any object that is big and red, or green. Those are artificial concepts with little to no relevance for the acquisition of real world concepts. Against that, Piaget’s work appeared irresistible: objects, quantity, causality, and more! That turned me towards Developmental Psychology. But still committed to experimentation with an unwarranted fascination for formalisms. I even envisaged formalizing Piaget’s theory of sensorimotor intelligence. Crazy idea.

Social Cognition? Having spent nine years in a boarding school, I must have had some latent interest in social relations. My first conscious memory of wondering about social interaction was much later when picking tobacco in Ontario; an illegal summer job for six weeks. It earned me a year’s Austrian university stipend. How, I wondered, could it be so difficult for five people, the farmer and us four pickers, to agree on how many leaves to clasp and take off the stem on each turn? I had no idea that I would soon come close to studying this very problem as a graduate student at University of Toronto.

Reading up on concepts, I realized that all experimental work was published in English, and I had better do something about that. What better than spend a year at an English-speaking university? I sent out about forty applications mostly to the USA and Canada, and a few to Australia. There were stipends to be had. Not much success. But VanderbiltFootnote ¹ offered me a fully funded place and the University of Toronto only a teaching assistantship. A passing American tenant in our flatshare persuaded me to avoid the conservative South and to opt for Toronto. So I did, despite less certain finances. It worked out in the end, with the province of Ontario coming to my fiscal rescue.

Off to Toronto. All set with a best value room booked at Rochdale College. Cheap flight to New York … bus to Toronto … Niagara Falls … border control. They took me in and questioned me for drugs, weapons, and so on. Why me? Never touched any drugs. Well! Rochdale was known for its hippie communes (remember Sharon Tate?) with its own security force (to keep the police at bay). I could prove my innocence: Rochdale was on the list of student residences in the official University of Toronto (U of T) student guide; and, fortunately, Greyhound and passengers took the delay in their stride.

In the 1970s the Department of Psychology at U of T was the powerhouse of memory research. Following fellow students’ advice, I opted for supervision by Anatol Rapoport, an unlikely choice for my interests but, in the end, a wise one. Anatol was a mathematician who had laid the game theoretic foundations for social sciences and was best known for his authoritative book on The Prisoner’s Dilemma. And since I had some maths background, he took me on. He quickly got me out of trying to formalize Piaget; formalization is of no use without deduction, he let me know. As it happened, John Flavell and Joachim Wohlwill had just proposed a little formal “Stage Competence Model” intended to differentiate Piagetian development in stages from usual learning curves. The Flavell-Wohlwill model predicted that correct answers on an easier version of a task should increase faster than on a more difficult variant. The lines diverge up to about 50 percent correct and then should converge before the easier version hits 100 percent correct. The learning model, in contrast, would predict divergence until each line reaches the ceiling. I set to work testing the models by teaching approximately six-year-olds simple set operations: union ∪, intersection ∩, subset ⊃, and so on. But the data provided no conclusive evidence one way or the other. The reason was sample size, that is, more revealing results would have needed hundreds of children. That was, in any event, my MA thesis.

After a brief peep into the possibility of studying with Adrien Pinard, a true Piagetian at l’Université de Québec á Montréal, I stayed with Anatol for my PhD, as his research assistant. We tested John Nash’s theory of the cooperative game. (You may know John from the film, A Beautiful Mind.) But for my thesis I stuck to children. As a compromise I worked on children’s decision-making under risk of uncertainty, including one study with a manipulable 2 × 2 game matrix. Each child had two carts and could push one forward to pick up goodies from the place where their carts met. I think that study was and remained the only one of its kind.

I sometimes wondered why I didn’t remain with game theory, an increasingly popular field with large swaths of uncharted developmental territory. Despite its elegant theoretical framework, however, I couldn’t quite see how to capture real social interaction. How would it help me understand our squabbles on the tobacco fields of Ontario? And there was my attraction to Piagetian concepts and increasingly also to artificial intelligence. This was strengthened when Heinz Wimmer, my old friend from Salzburg student days, asked me to join him writing a small German text, Kognitionspsychologie (modelled after Ulric Neisser’s groundbreaking Cognitive Psychology). Heinz took care of the language side. I contributed on object perception á la David Marr, explaining how our visual system uses sharp contrast lines to detect object edges and puts them together in a two-and-a-half-dimensional sketch of objects. Another chapter, for which I was responsible, was on action control in a GibsonianFootnote ² vein: “How hard to brake to get the car to stop just in front of the wall?”

These chapters were about how we process information into objects and actions. Game theory presumes many concepts as given: probability, subjective utility, payoff, choice, and so on. I guess this presumption of given objects was one reason for my losing interest in game theory and staying with issues of development à la Piaget and his concepts. In fact, there is a similarity to this distinction in how Piaget differs from so-called neo-Piagetians, such as Robbie Case, who I had met at David Olson’s regular meetings at Ontario Institute for Studies in Education (OISE). Neo-Piagetians have nothing to say about how concepts are acquired; only which tasks have too high an information-processing load to show conceptual competence. But concept acquisition was at the heart of the Piagetian enterprise. Thus, I found myself in complete agreement with Annette Karmiloff-Smith to call them – the Neo-Piagetians – anti-Piagetians.

After graduation I got a job in Basel, Switzerland. I had missed the Canadian immigration boat as the government had started to clamp down on immigration. But a job in Switzerland is not to be looked down upon. The only drawback with Switzerland was the size of the newly founded department, with only one full professor and one full-time plus one part-time assistant professor. So an offer from Experimental Psychology at Sussex University after a year came right in time to invest our riches of Swiss Francs in a house in Brighton and to enjoy the intellectual inspiration of an outstanding department. It had a great reputation for cognitive science and even artificial intelligence (AI), until AI became an independent discipline in the 1980s.

Heinz Wimmer rang again with the offer to work with him on his new grant from the Volkswagen Foundation on children’s stories of deception and counter-deception. Even AI had taken an interest in this, giving an analysis of the fairy tale Hänsel and Gretel (Bruce & Newman, 1978). Then we read Premack and Woodruff’s 1978 Behavioral and Brian Sciences paper, “Does the chimpanzee have a theory of mind?” In commentaries three philosophers recommended understanding of false beliefs as a test of that question. We realized that deception works by implanting a false belief, so we immediately switched from refining our deception stories to designing false belief tests (FBTs). And out came “Mistaken Max,” who hadn’t seen his mother put his chocolate in a different place and mistakenly thought it was still in its original place. Where will he look for his chocolate? Simple but effective (Wimmer and Perner, 1983). Max and his mother became famous in all parts of the world where psychologists started to investigate theory of mind. Under their pseudonyms Sally and Ann, the story characters became well known to children with autism who showed a specific deficit in attributing false beliefs (Baron-Cohen, Leslie, & Frith, 1985). The popularity of the FBT grew to the point that it became the “litmus test of theory of mind.” Why?

My suspicion is that the test captures two essential elements of a Theory of Mind (ToM), which is thought of as a theory about the mind for understanding behavior, that is, folk psychology. In the standard FBT two mental states are central, belief and desire. One has to understand that Max wants to get his chocolate and that he thinks (knows) he can get it by going to where it is. If Max knew where his chocolate was, one could predict his behavior (element no. 1) in a straightforward manner; he would go to where the chocolate is. The tricky part of the task is that Max is mistaken –for him his chocolate is still in its old place. So one has to take his erroneous perspective into account. Having a perspective is one (or the) characteristic of mental states. Hence, a child who manages to predict correctly must have understood something essential about the mind (element no. 2).

Understanding behavior is patently a fundamental evolutionary skill for any sentient organism but especially for humans, an extremely interactive species. So it seems to me the hype surrounding ToM was the discovery of folk psychology as something important and worth investigating. Why this insight came about 130 years after Psychology became a scientific discipline is an interesting puzzle; Bertram Malle (2006) says something about that. In any case, I now focus on the way this insight was used by different research traditions, a fascinating phenomenon for the sociologist of science.

One main strand in this domain focuses on the evolutionary importance of ToM. Evolution does not care much how an important ability is implemented. As long as you have it, your offspring will be selected for it. What is important is for it to be passed on and to function reliably. From this perspective a ToM (mentalist folk psychology) should be innate, as a highly automated domain-specific module, with no interference from other cognitive activity (hence a modular function). This nativist emphasis led to extensive efforts to find ever earlier evidence of a ToM in children. With modifications of the standard FBT, an impenetrable age barrier seemed to occur around three years, that is, there was no evidence for ToM any earlier. This held until (following our lead, Clements & Perner, 1994) nonverbal measures were used. Then the barrier dropped to thirteen months (Onishi & Baillargeon, 2005). Neural activity indicators lowered the barrier still further, to seven months. This trend ended a few years ago, in 2017, due to an increasing number of replication difficulties and evidence for alternative interpretations of existing data.

The other strand of ToM theory and research is committed to developmental change. Most of this work describes constituent concepts and their interrelation and provides a useful topographic map of developmental progress. Also, surprisingly few conditions have been identified that either accelerate or delay progress on the FBT. For example, two older siblings are worth one advanced year of development. In the other direction, language delay is the most striking factor affecting success on the FBT; it can make a difference of several years for hearing-impaired children. Also, factors usually affecting development, socioeconomic circumstances, and a child’s sex, have surprisingly minimal influence.

Nativist or not, neither of these traditions ventures beyond the extensively studied ToM domain. This is, I surmise, a wider consequence of the demise of the broad Piagetian framework, without which there are no theoretical grounds to expect any overarching domain effects in cognitive development. Moreover, much of the research leaves the fundamental questions untouched by not looking to alternatives.

Alternative Frameworks. Since the 1980s, philosophers (e.g., Gordon, 1986; Heal, 1986) have suggested that we do not entertain a theory of how the mind works but, rather, simulate how others’ minds work on our own mind. These discussions even referred to developmental data and received some attention in cognitive neuroscience. But in developmental research, apart from Paul Harris’s brief foray (1992) with simulation, this simulation alternative was largely ignored. How can that be? For it makes a big difference to what kind of skill children acquire whether they theorize or simulate. Let us look at Max and his chocolate again. For a theory children have to discover attribution rules like, “If someone doesn’t see an object moved, he believes it’s still in its old location,” or the so-called practical syllogism (as Gopnik and Meltzoff put it, 1997): “If a psychological agent wants event y and believes that action x will cause event y, he will do x.”

In contrast, if children simulate, they have to become adept at putting themselves in Max’s shoes (as the saying goes). They have to imagine being Max and simulate Max’s experience in the story: “I put my chocolate into the blue cupboard, go out to play, and then return to have some chocolate. Where is the chocolate? Where do I believe it is? Where will I go to get it?” To answer these questions, children have to become able to introspect (Goldman, 2006) on what they believe and intend to do or infer their belief via what Gordon (2007) refers to as an “ascent routine” – “Anything that is a fact I believe.” Finally, they have to leave their simulation and attribute the results to Max.

Why have such fundamental different approaches not been pursued? We did try to contribute to these fundamental issues; Toni Kühberger and I (2005) worked on a method to establish whether adults use theory or simulation to answer typical decision problems. But we did not receive much resonance in the literature. The topic simply seemed absent from anyone’s agenda.

Guided by philosophical action theory, Johannes Rössler and I took a sharp turn away from explaining all behavior with mental states via “teleology.” Ordinary explanations provide facts as reasons for an action, not mental states, for example, “Why did she start running?” “Cause the bus was coming.” Mental state explanations become indispensable when differences of perspective play a role as in the false belief task. This explains why children can explain actions long before passing the FBT and why they do not master the FBT before they pass other perspective tasks.

Cogworks. I have to own up. I was never much interested in the FBT as a litmus test of ToM, but rather in the cognitive challenge the test poses. It is one thing to acquire a concept of physical objects or events like house, island, or walking; quite another, it seems to me, to conceive of mental things like belief. An important observation for my theoretical pursuit was the strong correlation between passing the FBT and the ability to play an alternative naming game, such as: “We just called this a rose and also a flower. When puppet now says it’s a rose, you say it’s a flower, and the other way round” (Doherty & Perner, 1998). Domain specific theories can explain this correlation between tasks only by shunting responsibility to domain general changes in processing capacity or executive control. Whether this domain-general approach can deliver a satisfactory explanation of this and many similar correlations needs to be shown. Instead, I have developed a domain overarching explanation. For this I had to dig deeper into the cogworks of the mind.

On Mike Martin’s suggestion, I examined how discourse referents are analyzed in linguistics. These are mental representations of entities built up from text or discourse to enable reference to the same entity later in the text or discourse. A similar notion, object files, has been used in psychological theory and research on attention and object perception. A mental file represents an entity under a certain mode of presentation (= perspective). So co-referential files that represent the same entity are a general way to account for different perspectives; and here there arises a commonality between perspectives in the FBT and the alternative naming game. A different name defines a new perspective (Eve Clark, 1997). Via this line of reasoning we can account for why children who fail FBT also have difficulty with identity statements such as “The teacher is in the school. Susi’s aunt is the teacher. Where is Susi’s aunt?” Because co-referential expressions are involved, that is, “teacher” and “Susi’s aunt,” each triggers a file referring to the same referent.

Strong support for this theoretical position comes from the fact that, between three and five years of age, a host of additional tasks that require competence with co-referential files have been found to develop in step with the FBT: distinguishing appearance from reality, visual perspective taking, understanding photographic evidence, ambiguous figures, direction signs, and competitive games. These do not seem to belong to a single domain; or if they do, it is a quite different domain than hitherto conceived in psychology or cognitive science. In any event, I have compiled all relevant data and presented a detailed argument for this perspective in my forthcoming book, Mental Files in Perspective.

Now I must muster maximum executive control and curb my train of thought, or else we will land in narrow technicalities! But I hope to have given a sense of mental files as a promising part of the cogworks that go beneath the concepts we usually deploy to talk about the mind and its development. With such cogworks we can see commonalities across domains that would otherwise be inexplicable. And although mental files theory does not directly remind us of Piaget’s, it nevertheless captures his intuition that there are waves of general progress in development. Piaget called them stages, a concept which I had met in my first exciting encounter with the field of cognitive development. Now I have come back to this notion with an analysis in the spirit of contemporary cognitive science.

These, then, were the vicissitudes of influences that shaped the course of my intellectual preoccupations. I do sometimes wonder, though, what I would be writing about if I had been born earlier and met Konrad and his flock in person!

Final Reflections

Do I have any general advice for upcoming generations? Sounds presumptuous. But there has lately been much discussion about lack of replicability and of theoretical progress in psychology. This is not completely new though. I remember Endel Tulving, of episodic memory fame – must have been over thirty years ago – making this observation about research on human memory: despite 100 plus years of research since Ebbinghaus initiated the scientific field, we still do not have answers to his questions about the fundamentals of human memory. We get sidetracked investigating largely methodological issues in previous research and lose sight of the original questions.

I recently came across two interesting suggestions of how to counteract this tendency. The widespread confirmatory testing of isolated hypotheses has to give way to a focus on the proper derivation chain from theory to hypothesis to test (Scheel et al., 2020). And we should not routinely address questions within the currently entrenched framework. Instead, we should consciously engage in iterative, creative waves of working from the original phenomena towards conceptual clarity (Bringmann et al., 2022). I fully resonate with these suggestions. Theory of mind is a case in point (as I have suggested above). It has been uncritically adopted as the paradigm of how we explain behavior (folk psychology). Dominating the study of cognitive development for the last forty years, it has led to important discoveries about development. But it is now time to go back to basics and ask: Do we always mentalize about our own and others’ actions? An illuminating discussion can be found in Bob Gordon’s “Sellars’ Ryleans Revisited” (2000). What makes a psychological state mental? Do we really theorize, or do we simulate? Many more such foundational questions have been bypassed and remain unanswered.