#PredictiveProcessing

2024-11-10

Does anyone know about Helmholtz machines, or similar, and would be open to me asking a couple of questions?

I do experimental cognitive psychology and they're outside my expertise, and I think my questions are the type that are hard to answer through reading individual papers myself but should be simple for someone who is familiar with the higher level assumptions and norms of the field.

Thanks!

#academicChatter #predictiveProcessing #cognitiveScience #cogsci

2024-04-08

"Autistic and nonautistic adolescents do not differ in adaptation to gaze direction" doi.org/10.1002/aur.3118

I was reminded today that a talk I gave on this project for Neuromatch was recorded, so if you're interested in the paper but tldr, you can catch the main message in just 15 min 45 s here:

youtube.com/watch?v=v0Xn_VqE8r

#autism #cognitivePsychology #development #devSci #PredictiveProcessing

🤘 The Metal Dog 🤘TheMetalDog
2024-04-05



Music Evokes Distinct Bodily Sensations - Neuroscience News
A new study reveals how music evokes distinct bodily sensations, particularly in the heart and abdomen, linking these feelings to emotional responses and aesthetic appreciation.

neurosciencenews.com/music-bod










2024-03-27
2024-03-05

My paper on adaptation after-effects in autistic and non-autistic teenagers is out in Autism Research! 🎉

"Autistic and nonautistic adolescents do not differ in adaptation to gaze direction"

These autistic teens show a large adaptation after-effect behaviourally, though we don't see the after-effects in EEG, and we try to interpret this in light of Predictive Processing accounts of autism.

doi.org/10.1002/aur.3118

#autism #cognitivePsychology #development #devSci #PredictiveProcessing

Two panels with boxplots, the lefthand one subtitled "Pre-adaptation" and the righthand one subtitled "Post-adaptation". The legend shows blue solid lines for "Autistic group" and red dashed lines for "Nonautistic group". On both, the y axis reads "Proportion" and goes from 0 to 1 and the x axis reads "Adapted", "Direct" and "Unadapted". 

In the lefthand Pre-adaptation panel, the boxplots for both groups are around 0.25 for both Adapted and Unadapted, with Direct at around 0.5. In the righthand Post-adaptation panel, the boxplots for both groups are around 0 for Adapted, around 0.6 for Direct and around 0.3 for Unadapted. This means that both groups responded similarly to the manipulation and both showed a strong adaptation after-effect.
2024-03-04

“If so-called visual, tactile, or auditory sensory cortex is actually operating using a cascade of feedback from higher levels to actively predict the unfolding sensory signals… then we should not be surprised to find extensive multimodal and cross-modal effects… even on ‘early’ sensory response.” (Clark 2011) #perception #multisensory #CrossModalEffects #PredictiveProcessing

2024-03-04

“To predict the next word in a sentence, it helps to know stuff about grammar (and lots more, too). But one way to learn a surprising amount about grammar is just to look for the best ways to predict the next words in sentences. So you can use the prediction task to bootstrap your way to the grammar, which you then use in the prediction task in the future.” —Andy Clark (2011) #PredictiveProcessing #GenerativeAI #LLMs

2024-01-15

It's #ReadingMonday, and today I found an interesting paper on #predictiveprocessing models of the #brain

The paper reveals some of the processes in a modeler's mind that guide how we build models, and in particular predictive processing (PP) models.

  • The Bayes theorem is one thing, but the most interesting part when comparing PP models is how new hypotheses are challenged as they enter the generative model. In other words, what matters is how well the priors you include in your model fit the experiments.

  • To simplify, they identify two classes of priors, basically bottom-up and top-down, each with its pros and cons: "Cognitive-level models do not specify how they can be implemented in the brain and how the learning domains in these models can be learned. In contrast, neurobiological architecture-inspired models, although using neuronal-like architecture with learning starting from random weights, cannot account for important aspects of human cognition".

  • One underlying "meta-prior" is the existence of a hierarchy, which implies that there is both a feed-forward and a feed-back pathway, and that where they meet there are respectively "bottom-up or forward prediction errors" and "top-down or backward predictions". They conclude by suggesting that incorporating priors of both types may be a way to move the field forward.

2023-12-06

Very excited to share this preprint I wrote with Danaja Rutar, @LorijnZ , @francescopoli & Sabine Hunnius.

We first introduce #PredictiveProcessing and define its terms with *lots* of examples, and then point out that it cannot yet account for
#development

PP claims to be a unifying account of #cognition , and as such should apply to all humans.

We propose two additions which are necessary not only for completeness of PP,
but also for #devsci to be able to use it.

osf.io/preprints/psyarxiv/wktz

A set of graphs showing probability distributions over sizes of objects. There are 3 panels, A, B, and C, and each has a graph with a long caption.


A. Before I walk out of my house, I have a representation of the sizes of vehicles I have seen before, shown here in blue. Size is the variable we are depicting, and the expected value is represented by the probability of each value on the x axis. I have seen mostly cars of medium size, some smaller motorbikes and some larger vans, so the probabilities reflect this. The mean of this distribution is the point estimate of my prior expectation, and the inverse of the variance is the precision of the expectation.

B. When I do walk out of my house, I make a new observation. I see the elephant which has turned up on my street, and the estimate of its size is shown here in green. I have some uncertainty around the size of the elephant, because vision isn’t that reliable for estimating size, so the distribution has some variance. Because there is little overlap between the distributions, despite the uncertainty about the actual size, I can still tell that the elephant is larger than the expected vehicle size.
C. Under the standard Bayesian integration to update my model, I would come up with the new estimate of the value of the variable parked-object size, shown here in red. Note that this integration loses some of the specific details of the observation and results in a general expectation for the size of all parked objects.
pablolarahpablolarah
2023-09-21

☔️ The Faulty Weathermen of the Mind
by Shruti Ravindran @s_ravindran at @NautilusMag
Could a theory from the science of perception help crack the mysteries of psychosis?

nautil.us/the-faulty-weatherme

Silhouette and shadow of person's legs and half body walking in the night illuminated from behind by electric lights. The ground is wet with rain. Low angle take.
Image by StockSnap from Pixabay
Robert Roy Brittrobertroybritt@me.dm
2023-08-14

Nothing is real. Your every thought and action is affected by expectations brewed up in the nonconscious mind, the theory of #PredictiveProcessing contends. It aims to explain everything from a good golf swing to perceptions of chronic pain, even #consciousness itself. It has emerged from research in #philosophy, #psychiatry and #neuroscience. The big question: Can we exert control over the nonconscious brain and alter how we perceive everything and how we act and react? medium.com/wise-well/your-real

George Mussergmusser
2023-06-20

“The Dress was a really interesting case. Suddenly people realized there’s a lot of hidden variety in perceptual experience.… I think is going to ultimately give us something like a periodic table of experiential variation: all the neurotypical and the atypical cases and everything that lies in between, and all the variety within typical and atypical.”—from my interview with nautil.us/reality-is-your-brai

2023-06-13

#predictiveProcessing is a such a satisfying #neurological theory for how our brains work and my latest obsession.

Basically, our brain model lives in the future and sends predictions to our sensory organs and receive corrections instead of raw sensory data.

This explains why surprises are so satisfying (model gets to train itself) and exhausting (it takes energy to train).
Also maybe we can steer our model conciously to have a healthier #spiritual lives.
#optimism suddenly makes more sense.

Spektrum (inoffiziell)spektrum@anonsys.net
2023-04-02
Was ist die neuronale Signatur von Bewusstsein? Diese Frage beschäftigt Fachleute bis heute. Eine endgültige Antwort gibt es noch immer nicht – dafür aber eine Menge Theorien.
Infografik: Die wichtigsten Bewusstseinstheorien
KritischesDenkenPodcastKritischesDenkenPodcast@troet.cafe
2023-03-06

📢 Neue Episode des #kritischesDenken #Podcast!

Im Gespräch mit Prof. Philipp Sterzer über #Rationalität, #Irrationalität und wie #PredictiveProcessing uns dabei helfen kann, das alles zu begreifen.

Hört rein:
🎞️ youtu.be/ZsoZlEswsag

KritischesDenkenPodcastKritischesDenkenPodcast@troet.cafe
2023-02-12

Der Philosoph Prof. Tobias Schlicht im Kritisches Denken #Podcast:

Predictive Processing als ökonomische Arbeitsweise des Gehirns.

🎞️
youtube.com/watch?v=v4pNBT2MPl

#Bewusstsein #PredictiveProcessing
#kritischesDenken
#CriticalThinking

KritischesDenkenPodcastKritischesDenkenPodcast@troet.cafe
2023-02-04

Episode 72 vom #kritischesDenken #Podcast ist online!
Ein sehr interessantes Gespräch mit Prof. Tobias Schlicht von der
@ruhrunibochum
über das #Bewusstsein und #PredictiveProcessing.

🎧 kritisches-denken-podcast.de/e

#Philosophie #Interdisziplinarität

Client Info

Server: https://mastodon.social
Version: 2025.04
Repository: https://github.com/cyevgeniy/lmst