Through The Looking Glass: Towards a Neurobiology of Transference

When it comes to transference, can neuroscience inform therapy on whether it helps or hurts?

I’m in the woods and he’s climbing a tree, about to fall. I’m running towards him, reaching up to catch his flailing arm. Now he’s on the floor looking up at me, and I realize it’s my younger brother, looking just as he did when I was a teenager. But I thought I was running towards my five-year-old son; am I dreaming? It has to be... without permission, my brain did a substitution based on extracted similarities: the vulnerability, the need for protection, the panic of watching someone about to fall.

This happens when we’re awake too. We meet someone and something about them triggers a response we recognize from elsewhere. The new coworker whose laugh makes us inexplicably wary. The friend whose communication style has you walking on eggshells before we even know why. Our partner’s expression that pulls a reaction we know belongs to someone else entirely. The feeling and the reactivity arrive first. The analysis of whether it fits this particular person comes later, if at all.

All the above, and much more, fall broadly in what therapists have, and for a very long time, called transference. Although transference - the displacement of feelings, attitudes, beliefs, reactions - happens everywhere, all of the time, in therapy it has come to be regarded as one of the key instruments of the work.

But what happens if we add neuroscience into the mix. How might neuroscience conceptualize transference? Is it one or several neural processes? What could be the neural correlates of these processes? How could this knowledge help determine whether they are adaptive or maladaptive? And finally, can neuroscience inform novel ways of understanding it, working with it and ultimately ‘seeing through it’?

Because, you see, transference happens mostly out of awareness. If when we look into somebody, all we ever see is ourselves, we are looking at the mirror of our own experience. We don’t see the other, and life can become a series of self-fulfilling prophecies. Never changing. Never growing. So, to let go of transference is to get to the other side of the mirror of the self.

Once upon a time, answers to any of these questions would have been just a therapist’s wishful dream, far beyond the reach of what neuroscience could answer. Today however, we might be able to change our luck with some of them.

In this piece we will draw on neuro- and cognitive science in framing the problem of transference, for your consideration. Recent research in this domain has focused on tradeoffs people engage in to balance abstracting across experiences versus paying attention to the details of individual ones. This may be governed by how people deploy the variety of memory systems and executive control mechanisms each of our brains has. While the neurobiological details will be topics that we will continue to explore, in this piece we will focus on how the tradeoffs operate in relationships. To do so, we draw from empirical work on attachment representations, interpersonal accuracy and cognitive constraints. We will try to explain how these tradeoffs play out in therapy, how they relate to transference and when they are helpful versus not.

Here, we are just drawing the map of the territory. We have more questions than answers, and invite you to explore them with us…

Abstraction Revisited

A previous post discussed a fundamental tension in learning. When we encounter something new, each of our brains faces a choice: abstract structure that might transfer to other situations, or preserve the specific details of this particular experience.

Research on sequential learning revealed that, under specific conditions, people split into different strategies when facing this trade-off. Some rapidly extract shared structure across experiences, gaining the advantage of fast transfer but risking that new learning interferes with old memories. Others treat each experience as more distinct, protecting what they already know but potentially missing useful structure that repeats.

Neither approach wins universally. The optimal strategy depends on environmental structure. If structure genuinely repeats, extracting it accelerates learning. If contexts are truly distinct, maintaining separate representations prevents contamination. Abstract too much and you lose critical details, but preserve too much detail and you miss useful generalizations.

The same trade-off may operate in relationships. Some people might extract structure quickly across social interactions, learning fast but risking overgeneralization. Others may treat each relationship as fundamentally unique, protecting against misapplied expectations but potentially missing structure that would help them navigate recurring dynamics. The difference might not only be the level of abstraction we end up employing but also what we decide to generalize in other people’s behavior. Some of us might be better in extracting relevant patterns from other people’s behavior while others might do less well in this and therefore make inferences about other people’s behavior that do not match up. Furthermore, is this capacity something that is just immutable and fixed in each one of us or can we learn to do it differently and maybe even better?

These questions are in fact crucial for us as this process becomes especially visible in psychotherapy. A person sits across from their therapist and their brain immediately starts pattern matching. Authority figure. Caregiver. Person who will judge. Someone claiming to care who will eventually leave. The first person who’s actually listened. The therapeutic relationship offers a unique opportunity to observe this in real time, where these extracted structures may activate before one decides whether their therapist actually resembles whoever they’re being reminded of.

Transference as Abstraction

As briefly mentioned above, transference describes the unconscious process of transferring attitudes, beliefs, and feelings from one person (typically from the past) onto someone in the present. The classic therapeutic example involves seeing your therapist as a parent figure, responding to them with expectations, emotions, and behaviors shaped by that earlier relationship rather than by what’s actually happening in the room.

But the concept has evolved. Contemporary approaches recognize transference more broadly as applying structure extracted from past relational experiences to current situations that seem structurally similar. You’re not literally confusing your boss with your father. Your brain has extracted something like “authority figures who initially seem supportive but become critical,” a structure that may or may not actually describe your boss, but activates nonetheless.

Why does this happen? And is it always a bad thing? Of course not.

There is almost certainly something adaptive about this capacity. If every person required starting from zero, with no transfer from previous relationships, we’d be perpetually confused by basic human behavior. Our capacity to extract “this person avoids conflict like my sister did” or “this dynamic feels like my old boss” likely accelerates learning. We anticipate how interactions might unfold based on structural similarities to relationships we’ve already navigated.

Indeed, learning every situation as a unique instance requiring its own solution, would be computationally intractable. Making abstraction is one way some of the information can be reused without the additional cost of new learning or memory storage - we are effectively learning from past experiences.

And just to emphasize: None of this is abnormal, and frankly, it may not be something that can be turned off (maybe it’s the social equivalent of core object recognition in vision).

In The Therapy Room

Given that transference is an inevitability, what do we do about it in therapy? When we recognize it, do we think of it as a problem requiring a solution or a tool that can be used to accelerate the therapeutic process?

The answer is: it really depends and you can’t know in advance. Research on interpersonal accuracy suggests reliable individual differences in how well people make social judgments (Biesanz, 2010). Some people appear better at tracking which features of others’ behavior predict outcomes, maintaining “perceptive accuracy” across different relationships. Others struggle with what researchers term “perception spin,” where their judgments of others fluctuate in ways that reflect their own biases more than genuine changes in the other person’s behavior (Human & Biesanz, 2013). But even accuracy can critically depend on environmental structure; work on confirmation bias shows that holding strong beliefs and selectively weighting confirming evidence can actually be adaptive when those beliefs happen to match environmental dynamics and when people exhibit metacognitive insight into when they might be wrong (Rollwage & Fleming, 2021). The same process can be destructive when beliefs diverge from reality or when poor metacognition (can even be due to stress, instead of a trait) prevents updating (Beck, 1993). The difficulty in all of this is, we can’t really know in advance which is which.

Therapy puts this under direct observation. “You remind me of my mother” can be a breakthrough: the client realizes they’re bringing expectations from one relationship into another, can evaluate whether it fits, adjust accordingly. Other times the abstraction is the obstacle. The therapist isn’t actually like their mother, but the extracted similarity seems so strong that contrary evidence doesn’t register.

Also, transference runs both ways. Therapists have it too—countertransference—and it takes dedicated training to recognize and isolate. But the real power of the therapeutic relationship is that the client does in therapy what they do in life, only this time there’s someone outside their emotional sphere carefully observing.

Take a client who discloses only snippets of information based on what they think the therapist will focus on. They filter everything, believing their theory of mind is so good they know the reaction in advance. They regularly curate the narrative of their life. Over time, the therapist builds a completely wrong impression. The real life and the curated life become so far apart there’s an effective disconnect.

If the client stays long enough, once the therapist catches glimpses of what’s really happening, the real work can start. The work is about the story behind the story. Why the need to curate life this way? More likely than not, the client does this all the time, with most people in their life. It might be the root of considerable problems in relationships. Without even knowing it, this might be the real reason they came to therapy.

Working with transference means first bringing this pattern to awareness. Possibly trying to understand why it’s there, what function it provides. The work of therapy slowly brings to realization how what one believes to be accurate knowledge and pattern recognition of others is in fact an impediment to thriving, happiness and growth.

Individual Differences in Relational Learning

Why do some people see their therapist through the lens of everyone who came before, while others treat each relationship as starting from scratch? Why does one client immediately assume they know what you’re thinking while another struggles to extract any pattern at all?

As we mentioned above, the tension between abstraction and detail shows up in work discussed in a previous post. The details are not particularly critical (pun intended!), but the piece highlighted an experiment in which people learning sequential tasks split into two strategies: “lumpers” who extract shared structure rapidly (high transfer, high interference) and “splitters” who preserve distinct memories (low transfer, low interference). Neither strategy was universally better, indicating that the optimal approach depends on whether the behavioral strategy fit a particular environment.

The same trade-off might operate in relationships. Some people may extract structure quickly, transferring learning across relationships but risking overgeneralization. Others might treat each relationship as unique, protecting against misapplying old abstractions but missing useful structure. Most of us probably do both, in different domains and under different constraints.

Research on attachment provides evidence for these individual differences in how relational structure transfers across relationships. Attachment representations, formed through early caregiver relationships, function as relational schemas that guide behavior and expectations in new social situations (Pietromonaco & Barrett, 2006). Some work suggests that these schemas are organized hierarchically, with general working models applying broadly across relationships and more specific models applying to particular relationship types or even individual relationships (Baldwin, 1992; Overall et al., 2003). This general formulation is certainly congruent with the idea that hierarchical strategies are particularly efficient in navigating complex environments of which individual social networks are a special case.

The hardware each of us gets likely provides some constraints. Working memory capacity, for instance, affects how we process social information. Individual differences in working memory capacity predict performance across various cognitive tasks, including those requiring attention control in complex situations (Engle, 2002). People with higher working memory capacity show better ability to maintain flexible representations and resist automatic biases in person perception, while those with more limited capacity may rely more heavily on schematic processing.

The general idea is the following: Abstraction helps you learn fast but risks overgeneralization. Detail preservation keeps you accurate but slows transfer. Therapeutic work may happen in that space.

Where does this leave us? (And What’s Next)

Extraction of structure in relationships may be automatic and unavoidable. The brain faces a fundamental computational problem: extract shared features across social experiences to learn quickly, or preserve the specifics of each relationship to maintain accuracy. Neither approach dominates universally. The optimal strategy depends on whether the relational environment actually contains repeating structure worth abstracting, whether cognitive capacity supports maintaining flexible representations, and often whether this particular person happens to match previously extracted structure.

Transference in therapy puts this process under direct observation. The therapeutic relationship offers a unique opportunity to notice when extracted structure is being applied, to examine whether it fits, and potentially to develop flexibility in updating when it doesn’t. Our clinical examples illustrate how this plays out in practice: the client who curates their narrative based on anticipated reactions, effectively applying an extracted model of “how people respond” that may or may not match this particular therapist. The therapeutic work involves bringing this pattern into awareness, understanding its function, and recognizing when accurate pattern recognition has become an impediment.

The challenge is there’s no universally correct position on the abstraction-detail spectrum. Someone who aggressively extracts relational structure might thrive in an environment with genuine repeating dynamics but struggle when relationships are more varied. Someone who preserves excessive detail might maintain accuracy within individual relationships but fail to recognize recurring problems. Environmental structure matters as much as individual strategy.

What therapy might offer is not optimization of the trade-off itself, but awareness of when it’s operating. Recognizing “I’m seeing this person through the lens of someone else” doesn’t tell you whether the lens is accurate. But it creates space to wonder, to look for disconfirming evidence, to notice when expectations don’t match what’s actually happening. That flexibility, the capacity to hold both the extracted structure and the particulars of this specific person, might be more valuable than landing on any specific level of abstraction versus detail.

The framework also clarifies why therapeutic change can be difficult. Relationship schemas self-maintain through confirmation bias, selectively weighting evidence that fits and dismissing evidence that doesn’t. Breaking this cycle requires not just insight but repeated experiences that violate expectations strongly enough to update the underlying abstraction. The therapeutic relationship, with its consistency and safety, may provide conditions where such updating becomes possible.

But we are left with more questions than answers. For example, what are the set of neural processes that map onto what we call transference? In other words, are the processes behind learning the abstraction of what another person is likely to be towards us the same as the process behind the abstraction of what this person thinks and how they behave, in general? What is the role of emotions in this process? How much of all of this overlaps with the process that is linked to what is referred to as ‘theory of mind’ or ‘mentalizing’ network? And finally, are there any parts of these processes that we can volitionally control or influence? In other words, is there something we could do in therapy that would make working with transference more effective than we already do?

And maybe, just maybe we have taken you one step closer to knowing how to, like Alice, get to the other side of that transference looking glass.

---

Thank you for reading. Mike and Ana would love to hear your thoughts on this framework. Does it match your experience? Does it help explain patterns you’ve noticed in your own relationships or therapeutic work? If this is interesting to you, please consider subscribing and sharing with your network.

---

References

Baldwin, M. W. (1992). Relational schemas and the processing of social information. Psychological Bulletin, 112(3), 461-484.

Beck, A. T. (1993). Cognitive therapy: Past, present, and future. Journal of Consulting and Clinical Psychology, 61(2), 194-198.

Biesanz, J. C. (2010). The social accuracy model of interpersonal perception: Assessing individual differences in perceptive and expressive accuracy. Multivariate Behavioral Research, 45(5), 853-885.

Engle, R. W. (2002). Working memory capacity as executive attention. Current Directions in Psychological Science, 11(1), 19-23.

Human, L. J., & Biesanz, J. C. (2013). Targeting the good target: An integrative review of the characteristics and consequences of being accurately perceived. Personality and Social Psychology Review, 17(3), 248-272.

Overall, N. C., Fletcher, G. J., & Friesen, M. D. (2003). Mapping the intimate relationship mind: Comparisons between three models of attachment representations. Personality and Social Psychology Bulletin, 29(12), 1479-1493.

Pietromonaco, P. R., & Barrett, L. F. (2006). Internal working models: What do we really know about the self in relation to others? Review of General Psychology, 10(2), 155-175.

Rollwage, M., & Fleming, S. M. (2021). Confirmation bias is adaptive when coupled with efficient metacognition. Philosophical Transactions of the Royal Society B, 376(1822), 20200131.

Weiner, I. B., & Bornstein, R. F. (2009). Principles of psychotherapy: Promoting evidence-based psychodynamic practice (3rd ed.). John Wiley & Sons, Inc..