Reading and predicting thoughts and behaviours

[Cara]

I’m sure many have noticed that recently there is more overt/open media coverage of the development and implementation of tools and capabilities that aim to predict and characterise individual behaviour and thoughts. These are mostly presented as well-meaning projects in service of an early identification - early remediation approach.

In some ways this is not so different from psychometric tests. However, the intent of these projects seems to creep ever more towards the idea of “pre-crime” prediction. There are obvious echoes of phrenology (https://www.britannica.com/topic/phrenology) and eugenics (https://www.britannica.com/science/eugenics-genetics).

There seem to be two general approaches to the topic:
- algorithmic modelling of behaviour based on behavioural tracking, big data, and AI modelling / prediction (e.g. speculations about Palantir)
- “medical” scanning of individual brainwave patterns, brain structures and other physiological or anatomical characteristics of individuals to (a) build a reference pattern, and (b) then compare and contrast future “suspects” to this derived pattern.

In these guises the projects seem unhappily dystopian and to be ever more like the scenario shown in the film Minority Report.

This thread is to gather reports of academic articles and studies that seem to be focussed on the development of these capabilities.


Please do add any articles, links etc. that cover studies and projects of this kind.

[Cara]

This article covers brain scanning done on prisoners that compares the brain structures across groups classified according to level of violence involved in the crime.

The author does suggest ethical and moral issues but diminishes them in the language he conveys these concerns in (“it is fair to be mildly unsettled”).

Another problem is the not so subtle inference (not outright stated) that murderers are less intelligent (implied by the phrase “significant grey matter reductions”).

https://newatlas.com/brain-scan-murd...science/60510/

Inside the head of a killer: Imaging study uncovers unique brain abnormalities in murderers
Rich Haridy

For decades scientists have been exploring the minds of criminals, trying to understand what makes a person commit a violent, heinous act. The advent of modern brain imaging techniques in the late 20th century proved a boon to the burgeoning field of neurocriminology. Examining the brains of murderers, psychopaths and other violent offenders has unsurprisingly proved to be a controversial scientific pursuit. To some, modern attempts to associate criminal behavior with objective brain imaging skirts discomfortingly close to the controversial 19th century pseudoscience of phrenology.

Phrenology asserted that the shape and contours of a person's skull could be directly associated with a variety of personality traits, from one's intelligence to their propensity for criminal activities. Although the "science" of phrenology was pretty widely discredited by the mid-1800s, its central notion has lingered as a supposed scientific basis for many racist ideas into the 20th century.

Most modern neuroscience looking to find structural or biological signs of criminal and antisocial behaviors in brain imaging data concentrates on more general traits such as aggression or lack of empathy. The neurology of psychopathy is a rich field of neuroscience, with many researchers uncovering fascinating insights into potential brain structures that could underlie certain violent antisocial behaviors.

A new study, published in the journal Brain Imaging and Behavior, is taking this idea of neuroimaging criminal behavior one step further than any prior research. This new research, based on brain imaging data from 808 incarcerated male adults, suggests the brains of individuals that have committed murder are notably different from other criminals, both violent and non-violent.

Specifically imaging the brains of homicide offenders in not especially novel. Pioneering neurocriminologist Adrian Raine conducted some of the earliest neuroimaging studies on murderers in the 1990s, homing in on several brain regions that seemed to be related to homicidal behaviors. However, as corresponding author on this new study Kent Kiehl tells New Atlas, much of this prior neuroimaging work commingled homicidal behavior with other psychiatric conditions such as schizophrenia or psychopathy.

"These early studies relied on murderers who had been found not guilty by reason of insanity (NGRI), and thus include effects of comorbid psychosis and organic brain injury alongside those effects specifically related to homicidal behavior," Kiehl tells us via email. "So yes, our study is novel in that we excluded men with psychosis and brain injury. We also studied nearly 1,000 total offenders in this paper. By far the largest study to date."

The research looked at structural MRI data from subjects categorized into three groups: 203 "homicidal" subjects (including self-reported homicide offenses and explicit attempted murder convicts), 475 violent but non-homicidal subjects (including assault, domestic violence, and other cases consisting of serious bodily harm), and 130 minimally violent subjects (drug possession, prostitution and other crimes resulting in no serious injury to others).

The results strikingly found a number of gray matter differences in the brains of those subjects who committed a homicide, compared to the other two groups. In fact, the researchers suggest there was little difference between the violent but non-homicidal subjects and the minimally violent subjects, implying there are significant gray matter reductions in several brain areas that particularly distinguish homicide offenders from other kinds of criminals.

"The orbital frontal cortex and anterior temporal lobes showed the largest effect sizes; that is, men who committed homicide had less gray matter in these regions than other violent or nonviolent offenders," Kiehl tells us.

These results are arguably rather strange considering the heterogeneous nature of homicide. The study does note it only included serious homicide offenses in the homicide cohort. So accomplices to murder were not included, nor were possible accidental death cases. However, this still strangely mashes up over 200 homicide offenders into a single group. Intent, for example, is not clearly accounted for here, so the study implies little neurological difference between a homicide that resulted from an aggravated assault gone wrong and a more violent, intentional act of murder.

The study does clearly note, "it should not be mistaken for the ability to identify individual homicide offenders using brain data alone, nor should this work be interpreted as predicting future homicidal behavior."

Although, directly questioning Kiehl on this point, he suggests predicting future homicidal behavior from brain data is a reasonable future outcome from this kind of work … and the more data that can be gathered, the more accurate these predictions may ultimately be.

"Yes, this is a first step towards using neuroscience to help predict who will commit homicidal behavior and identifying neuro-risk factors for homicidal behavior," Kiehl tells us. "Our future goals include following up large samples of high risk youth to see if the regions we have identified in this study do predict future homicidal behavior in those samples."

It is fair to be mildly unsettled at the ultimate implications of this research. Kiehl is gesturing to a degree of biological determinism here, suggesting there could be a specific brain "fingerprint" that identifies a person with the capacity to commit murder. Does this mean brain imaging could become major evidence in future murder trials? Or even more disturbing is the question of what we do with the knowledge that there may be an identifiable kind of homicidal brain?

Neurocriminologist Adrian Raine suggests these kinds of criminal brain fingerprints don't necessarily predict future antisocial behaviors but instead imply a certain greater propensity for these kinds of behaviors. When asked in 2013 whether he would be comfortable subjecting his own children to a brain scan that could preemptively assess violent tendencies, he likens the hypothetical process to a kind of public health crime prevention program, where behaviors may be identified and changed at young ages before they develop.

"If there was the opportunity for screening at school or through a GP programme, would I do it? Well, if my kids had problems, as a parent I would want to know about them and I would want to know how I might deal with them," Raine told The Guardian in 2013. "If you brought in such things as emotion regulation and impulse control, which we know are risk factors for behavior, then to me, as a parent, I would sort of want to know what could be done to help with those."

Kiehl is perhaps a little more reticent to hypothesize on the social or cultural outcomes of his line of research. It may be evasive, or merely the classically cold, blinkered vision of a scientist, and Kiehl accepts the potential directions his research could lead in but simply asserts more work is needed before any confident conclusions can be made.

"This study definitely points in the direction of developing a brain 'fingerprint' for homicidal behavior, but our results need to be replicated and we need to conduct longitudinal studies in order to more fully address this critical issue," Kiehl concludes.

The new study was published in the journal Brain Imaging and Behavior.

[amor]

Why not scan the brains of the World Leaders responsible for the Canadian Genocide of First Nation Indians now occupying mass graves, the Bankers who enslave us by turning our hard labor, via FIAT currency, into legal corporations and city buildings which they own and we cannot and call it Organized Theft. I am certain that a difference will be found. They may be smarter than we are but their Brains will represent a CRIMINAL PROFILE.

[Cara]

Here’s another project in this vein. This article is from 2017.

https://www.popularmechanics.com/sci...th-brain-scan/

Scientists Can Now Read Your Thoughts With a Brain Scan
Researchers at Carnegie Mellon have made headway decoding the "alphabet" of the brain.

By Avery Thompson, Jun 27, 2017

Your innermost thoughts might not be so secret after all. At least, if you're thinking them inside an MRI machine.

Researchers at Carnegie Mellon University developed an algorithm that can decode a person's thoughts using data from a brain scan. The researchers used this algorithm to gain insights into how our brains work and form complex topics.

Our thoughts can range from simple to highly complex. A simple thought might just consist of a single concept, like "store," or a more complex group of concepts, like "I went to the store last Saturday in to the store last Saturday in the rain." Research shows that the brain breaks down these complex thoughts into smaller pieces, each corresponding to a different aspect of the thought.

Essentially, the brain uses an alphabet of around 42 different elements, each referring to a specific concept like size, color, or location. The brain combines those together to form complex thoughts.

Each of the "letters" in the brain's alphabet is handled by a different part of the brain, so by studying brain activity with an MRI machine it's possible to determine what a person is thinking about.

The researchers developed an algorithm to do exactly that, though they didn't put it to work literally reading minds. Instead, they gave it MRI scans and their corresponding thoughts, and the algorithm would then predict what a completely different thought would look like in the MRI. In the study, it managed to do this with an 87 percent accuracy. And while this tech was not used to attempt to read actual minds, it seems likely that it could.

Nobody is going to use this tech to secretly read your mind anytime soon. But neuroscientists should be excited about this glimpse into the inner workings of our own brains. This study goes a long way toward figuring out how we think, which helps to solve one of the many mysteries of our own minds.

Source: Carnegie Mellon University

Here is the title, authors and abstract of the paper:

Predicting the Brain Activation Pattern Associated With the Propositional Content of a Sentence: Modeling Neural Representations of Events and States
Jing Wang, Vladimir L. Cherkassky, and Marcel Adam Just *
Center for Cognitive Brain Imaging, Psychology Department, Carnegie Mellon University, Pittsburgh, Pennsylvania

Abstract:
Even though much has recently been learned about the neural representation of individual concepts and categories, neuroimaging research is only beginning to reveal how more complex thoughts, such as event and state descriptions, are neurally represented. We present a predictive computational theory of the neural representations of individual events and states as they are described in 240 sentences. Regression models were trained to determine the mapping between 42 neurally plausible semantic features (NPSFs) and thematic roles of the concepts of a proposition and the fMRI activation patterns of various cortical regions that process different types of information. Given a semantic characterization of the content of a sentence that is new to the model, the model can reliably predict the resulting neural signature, or, given an observed neural signature of a new sentence, the model can predict its semantic content. The models were also reliably generalizable across participants. This computational model provides an account of the brain representation of a complex yet fundamental unit of thought, namely, the conceptual content of a proposition. In addition to characterizing a sen- tence representation at the level of the semantic and thematic features of its component concepts, factor analysis was used to develop a higher level characterization of a sentence, specifying the general type of event representation that the sentence evokes (e.g., a social interaction versus a change of physical state) and the voxel locations most strongly associated with each of the factors. Hum Brain Mapp 00:000–000, 2017.

Full paper here: http://www.ccbi.cmu.edu/reprints/Wan...l-preprint.pdf

~~~

The study combined MRI brain scans with a computational model (algorithm) that used concepts from linguistic analysis. So we have the combination of electromagnetics, computational and statistical modelling, and linguistics theory.