Scene from Black Mirror’s episode Crocodile, where this woman can see the other’s memories through a screen

Mind-reading technologies could break with anthropocentrism

A recent study used Stable Diffusion to reconstruct images shown to patients. Its accuracy has led scientists to question whether the same tools could be used to also understand other species’ perceptions of the world.

Lidia Zuin
5 min readMar 28, 2023


You know that expression seeing with someone else’s eyes? The fact is that we are getting closer to making this possible with the help of artificial intelligence (AI) and brain-computer interfaces (BCI).

Last week, I briefly mentioned a study that used the image generation algorithm Stable Diffusion to reconstruct images seen by subjects who had their brain scanned while exposed to these pictures.

Unprecedentedly, the surprising accuracy of the algorithm led scientists to wonder whether, in the future, these same tools could be used to access people’s dreams and thoughts or even to understand how other species perceive the world.

For now, this process is not performed in real time, as seen in movies such as Spielberg’s AI or in Black Mirror’s episode Crocodile, but after the analysis of scan results — similarly to the interpretation of MRI results, for example. But, on the other hand, there are other techniques such as deep brain stimulation which aim for a more simultaneous reading, so that the technology could help people (for instance, those who suffer from multiple sclerosis) to communicate.

Also, the way we communicate with the world and how we express our ideas and feelings are some of the most relevant topics in disciplines such as semiotics. In the case of cultural semiotics, the theorist Ivan Bystrina suggested categorizing these communicational manifestations in units he called “text”, though that would not be limited to a written format.

During a lecture presented in São Paulo in 1995, Bystrina presented the concept of text as a group of signs/symbols which may have communicational and informative, expressive, emotive, aesthetic, and social values. The researcher suggested a division in three categories:

  • Instrumental texts are those related to our day to day life, they are extremely pragmatic and technical (e.g. instructions manuals).
  • Rational texts are related to natural sciences, to mathematics (e.g. scientific papers about artificial intelligence).
  • Creative and imaginative texts include myths, rites, ideologies, fictions and artwork.

This same interpretation is also seen in the book Emotion and Imagination by Norval Baitello Jr. and Christoph Wulf. Here the authors suggest the term “image” instead of “text” as a means to refer not only to visual, but also auditive, olfactive or even proprioceptive (that is, the internal/mental image we have of ourselves) images.

When an artist transfers this mental image to the physical plan, for instance, through a sculpture or a painting, they can do this in several different ways. Think, for example, how Caravaggio’s self portrait is different from Van Gogh’s not simply on a technical or stylistic level, but also because they communicate different ways to understand and see the world. Or take the nautic landscapes painted by William Turner and how they weren’t necessarily picturing a real storm in the ocean, but rather a mental state.

This relationship between the idea and the work gets even more complicated when we consider surrealist, abstract, or dadaist art, for instance. It is even “worse” when we include contemporary art in this conversation, since, many times, audiences can think it is about a presumption of an idea rather than an actual invitation to discussion due to its displacement of technique, aesthetic and communicative referential (which makes me think of that famous Magritte painting which figures a pipe and the description “this is not a pipe”).

Above, the original picture seen by the patients and below the reconstruction made by AI

With all these things in mind, I wonder how accurate these “mind-reading” technologies really are. As seen in the study I mentioned in the beginning, despite the accuracy of this new algorithm, we can still see discrepancies between the original picture seen by the patient and the reconstruction performed by the program. Though both images look like a bear plushie, they are different bear plushies, even in color. What is surprising though is that the perspective and the composition of the images are usually precisely replicated.

For now, this technical achievement does not really take us to a loyal projection of our mind images. It is just the same way when you imagine a stickman and how different it will look like when you draw it on a paper, not simply due to technical and material constraints, but also because there is an abstraction included to that mental image which is also contaminated by emotions and memory. In other words, even if it would be possible to take a “screenshot” of our brain, there is much more context to be considered to understand that image than what the reconstruction could show.

This is especially important when we think, for example, about the risks that mind-reading technologies pose not only to privacy, but also its impacts on health. That’s why discussions about neurorights are already taking over the constitution of countries like Chile, one of the first to join that conversation.

On the other hand, a more dystopian one, this kind of intervention can lead to exploitative breaches as those envisioned by Jonathan Crary in 24/7: Late capitalism and the ends of sleep. In fact, there is already research on how to insert specific contents to someone’s dream, especially ads. That’s why 40 sleep researchers signed an open letter pleading for the regulamentation of what they call targeted dream incubation or TDI. According to them, companies like Coors’ and Burger King are already actively working on this.

As we inherit the idea (or hope) that our mind is the ultimate resource for privacy, we resist the fact that new technologies could break with that notion. On the other hand, I wonder how much these biological discoveries ignited by digital technologies could affect philosophical disciplines such as epistemology (study of thinking) and ontology (study of being).

If we are really just cucumbers with anxiety, as the joke says, there is a great possibility that we change the way we perceive our world and specially the way we relate to other species. Up to now, our anthropocentric approach insists on putting humanity on another existential level due to the way our mind was shaped, but maybe we are simply (very) complex organisms that ultimately may have their code cracked.

Extra resources:

Disclaimer: This essay was originally published at Tilt UOL. This is a translation from the Portuguese version.

Did you like this story? You can show support by buying me a coffee. :)



Lidia Zuin

Brazilian journalist, MA in Semiotics and PhD in Visual Arts. Researcher and essayist. Technical and science fiction writer.