In an app I am creating I am profiling user behavior in a multitude of different ways. One of the techniques I would like to use is track the imagery they interact with to get a handle on their current mood. The imagery uses an internal tagging nomenclature that indicates various aspects of the image one of which is the key feeling it is intended to trigger. The feelings I am using are something along the lines of

Health, Desire, Hunger, Comfort, Indulgence, Relief, Peace, Relaxation, Nurture, Creativity, Excitement

To provide a context here is a clutch of images (not the actual ones I am using)

enter image description here

My intent here is to attempt to build up a corpus of knowledge regarding the moods of the user and the way they vary over time and space based on what types of imagery they interact - in the sense of tap on to engage with the asssociated content - with.

I use a large range of profiling techniques most of which have a firm basis in published research. And then every once in a while I think of an idea such as this one "all on my own". Which brings me to my question - is there any scientific literature on the subject that reduces feelings triggered by images to a core list? Having such a list would assist me in tagging the imagery I use in the app.

I have researched the subject on Google, ResearchGate and Elsevier but come up with nothing useful. This could be in part because I am searching for the wrong terms - I am afterall, a mere IT guy who is blundering into psychology.

I have found useful information in resources such as this one. Whilst I appreciate the insights and sense that they are talking sense there is very little by way of referenced research to back up what is claimed.

  • $\begingroup$ The golden standard is free response (i.e., letting participants rate their own moods with their own labels). Forced-choice lists can limit the validity of your data. If the free-response format is too high dimensional for you, then you can use something like the PANAS with an "Other - Please fill in blank" as needed. $\endgroup$ – mrt Jun 21 '18 at 19:14
  • $\begingroup$ Thank you. In the present context though the option of asking app users to identify the emotion (I prefer to think evocation rather than emotion) that made them tap on a certain image is not there. Rather it is for the app to record - amongst other things the associated evocation tag on the image with a view to building up a bigger picture of the user's behavior. $\endgroup$ – DroidOS Jun 22 '18 at 9:54

I am pleased to see the interest that this question has aroused. I have eventually managed to find one scientific article on the subject - http://www.imageemotion.org/machajdik_hanbury_affective_image_classification.pdf.

which provides me with a decent starting point. I would be most grateful to pointers to any other research in the field.

  • $\begingroup$ Nice find. I had no idea such research existed. The way you find more of the same is (1) read the paper's related research, e.g. "In [30], analysis and retrieval of images at the affective level is called Emotional Semantic Image Retrieval (ESIR)". That should cover most relevant prior research. And (2) use Google Scholar (or similar databases ACM's might be good in this case) to find newer research citing this paper. $\endgroup$ – Fizz Jul 1 '18 at 14:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.