In this study we run three web experiment asking how we perceive 100 common feelings by assessing them i) over 5 dimensional scales, ii) similarity between them iii) how they map on the body and iv) collected similar terms from @neurosynth 2/7
Results showed 5 robust clusters with DBSCAN and visualization with tSNE gave clear components related to emotional valence as well as mental saliency. 3/7
Here mapped over the bodies 4/7
And finally we assesses the similarity of the representations of the various domains we tested as well as the semantic distance between the words and how they are represented as neurosynth terms. 5/7
I conclude the thread with @aivoAALTO 's words "We have obtained solid evidence that shows the body is involved in all types of cognitive and emotional functions. In other words, the human mind is strongly embodied." 6/7
Bonus: all data and code are fully available zenodo.org/record/1291729 And for the sake of transparent #openscience, even the full GIT history! With first commit dating March 2016 (mostly me fighting with CSS to get the web experiments right; no bad words found in GIT history). 7/7
Try one of the experiments translated into English here emotion.nbe.aalto.fi/feelings_sim/ by arranging the 100 words inside the box (the closer the words, the more similar they feel to you).
Since @Twitter 's re-encoding of the video might lose some details, high quality version of the video with the maps can be seen here: onni.utu.fi/maps_of_feelin…
This is important. Yes, re-identification of de-identified brain MRIs is possible, but it is considered to be a crime under GDPR, implemented as “data protection” laws in EU (each country made their own laws and legal consequences).
Pseudo-anonymisation assumes that it is “highly unlikely” to re-identify the data. This means that for those cases in which it becomes trivial to reidentify a subject (eg rare case study subject), special care is needed and it should be the ethical committee with the researcher..
...to decide on benefit for science/society vs potential harm to the individual. Also: I believe that sensitive data can still be shared on a per request basis as done by human connectome project or ADNI, but legislators have still to finish their job as these relate...
As somebody pointed out already, pyramid schemes are everywhere outside academia. I’d love to work as a (data) scientist for a non-profit NGO, but as soon as profits are involved I start to question the ethic of my work. 1/n
Academia is still one of the most ethical options (if done with integrity and transparency) as I feel I am putting my skills for the benefit of society, even if basic scientific research cannot be directly translated into applications. 2/n
In my opinion the issue is that the system supports only the rockstar PI culture: there are basically no permanent positions for staff scientists. You can only become a PI/professor or leave. 3/n
I’m a Matlab “guru” (or so they say) since 1999, but I’ve decided to switch to Python as the future for me can only be there. Here’s a nice book to do the switch enthought.com/white-paper-ma… 1/n
Don’t get me wrong, I love Matlab. I’ve done anything with Matlab from signal processing toolboxes to art installations (!). 2/n
I’ve made music with Matlab and beautiful scientific pictures. But I feel like a scientist from 100 years ago who was publishing articles in German (or French or Italian). All languages are beautiful and can be used to express anything... 3/n
Excellent educative talk based on measures from structural MRI and their correlations with for example age. See ncbi.nlm.nih.gov/pubmed/27103141
Chris @cMadan is now talking about fractal dimensionality: a measure how complex a structure is (a sphere has low fractal dim.) which also correlates with age.
Brainhack Global 18 in Finland has just started with prof. Em Riitta Hari @aivoAALTO talking about the unsolved questions in neuroscience #bhg18#bhg18fi
"how good are your data?" often theory comes before data (e.g. physics) and many labs do same experiment and results are combined #bhg18#bhg18fi
"animal models, are we using the best species out of 8M species in the planet?" #bhg18#bhg18fi