Indirect mood measures


I was wondering if anyone has tried less direct measures of mood. There are so many challenges with tracking mood and a big one seems to me the fact that tracking or thinking about mood in itself affects mood.
This wouldn’t completely get around it, but one thought I had was to choose a variety of songs to put on a ipod shuffle and at a few random times in the day take a five-minute break or so and record how long until I skipped for any given song. There are some obvious predictions, like I’d be more likely to dwell on a sad song when I’m sad, but I feel like there’s much more subtlety than that with changes in attention and interest in reflection that accompany different moods. So I guess my first step would just be exploratory and if it seems like a valid or interesting measure, I might use it and test out some correlations with other basic life stats for a while.
What do you think?

As a side note, I was poking around this mood forum last night and one thing I’ve been mulling over is the connection between mindfulness and tracking that I think a couple people suggested. The two strike me as almost antithetical-- of the many virtues of recording and analyzing life stats, focusing on the present moment doesn’t seem to be one of them!
I’d love to hear your thoughts on that too.


1 Like

Hi Jane,

I really like your idea of using music to indirectly track your mood! Might be hard to build that kind of system, but I’m sure someone could :slight_smile:

For another option you can always try Ian Li’s Mood Jam. It only asks you to pick colors instead of using word (although you can annotate). Ian is a good friend of QS (and past organizer for QS Pittsburg). had a mood report. Don’t know if it’s still working, or if this data is accessible through their API.

Hi Jane,

I’ve tried less direct measures of mood - including how much time I spend meditating per day, and how many journal entries I write per day.

These aren’t direct measures, and they don’t work over the short term. You can’t tell anything without a few months of data, but over the space of months and years I’ve found they have given a very powerful insight into how my mood has changed. There are some obvious features - e.g. when I’m depressed, I don’t meditate, so I can see my depressions by looking for these troughs in my data, as well as some longer term patterns that I can relate to changes in my mood.

Over the shorter term I have found that recording something simple once a day can be effective - e.g. “did I feel depressed?”, “did I feel disturbed?”. This takes a bit of care and subtlety to do right, and if you have a mood disorder or depression it can lead to many days in a row of simply saying “yes”. However, this makes the “no” days much more interesting if and when they occur, and even more so if a pattern emerges. I’ve found it helpful not to expect to learn anything from this for at least a month, and to be ok with boring data.

I think this kind of daily check-in relates to mindfulness - not in the sense of being constantly in the present, but in the sense of directing your attention in a non-evaluative manner to some aspect of yourself and simply observing.

My experience is that repeating this over a period of time - a couple of months or longer, has helped me to develop more intuition about my mood, and also more patience for times when I’m not doing so well - because I can more easily remember that it’s not always like that.

I think the music tracking idea is a great one too - and I would imagine that over time you’d see some interesting patterns. I’d suggest that you’ll probably find something interesting in the more ‘gross’ features of the data - e.g. how many times per day you do it at all, what days do you not engage in the practice, as much as you will in the moment by moment tracking.



You may want to consider various physiological functions to indirectly measure mood, such as blood pressure, heart rate, respiratory rate, skin temperature and brain waves.

Following on Joshua’s point about physiological measures, I’ve seen some interesting things about heart rate variability as I’ve started exploring quantified self. A talk by Stephen Jonas was particularly interesting He used a heart rate variability monitor to not only track mood, but also to give him an alert whenever the software detected he was too stressed. He also says at some point something like that it made him more mindful, and there were times when he figured out that he was stressed just before the monitor did. Maybe this process can encourage mindfulness, by training the brain to pay attention to signals of stress? I’ve also seen a suggestion that heart rate variability could be linked to willpower/self-regulation, which one might expect to come out of greater mindfulness

I’m really intrigued by these ideas, although I don’t know if I’ll follow up on them any time soon. I have started looking for technology that might be able to provide a useful measure of HRV (particularly providing alerts) in a less disruptive and broader way. Although it doesn’t provide alerts, the W/Me on kickstarter looks promising. They claim to measure four mood states, but if it’s all HRV I have no idea how accurate that would be. There are also other heart rate chest straps and watches that might be able to do something similar and provide alerts with a little hacking (and a little validation of how accurate they are).


Great point about the lack of validation as the current norm for most consumer tracking devices. We recently had Jonny Farringdon speak at an event we co-produced. He is the director of informatics at BodyMedia and he noted how their devices are FDA approved and thus have had to undergo much more extensive testing compared to products such as Fitbit, Nike Fuel, etc. Personally I don’t think have more devices go through FDA approved is the way to go, but I think consumers would benefit from some industry standards, etc. What do you think?

Link to Jonny’s talk:

I think indirect measures are really interesting, as they probably will help tease out many different components of “mood,” both psychological/physiological and contextual. I use “episodes of anger” as an indirect measure. I simply make a note in my tracking spreadsheet whenever I lose my temper. When this happens, I take it as a signal of poor emotional condition. That is: bad mood! The nice thing about this signal is it is clearly noticeable, unambiguous, and rare enough (less than 1x per week usually, but more when I’m not doing well) to be meaningful.


Interesting talk. I didn’t know how widely the Bodymedia line of devices was used or how well they were tested. I think validation in quantified self can probably involve figuring out lightweight but useful methods. For heart rate variablity, I had two kinds of ideas: 1) measure the device relative to a known accurate device or 2) validate that device compared to the states you want to track. 2) could have different experimental forms, like a random reminder to record a subjective evaluation of your current state, and then later on compare that data to what the device reports at those times.

First post - fascinating website - where has google been hiding you all these years.

I have been trying to work out an indirect measure system for some time now to manage my bi-polar. I have tried / used various gadgets and methods to self-asses over the years but have personally found the data is not directly useful in a point of time context - but sometimes more helpful aggregated over a longer time. Primarily this is because I need to be in a particular state of mind to capture the point in time data in the first place. Secondly - it is interesting to see how your overal state in a particular day,week,month,year cycle skews your self assessments. I have identified cycles that are time based - to put simply - say when in a generally good cycle over a month; daily / weekly self assessment skews upwards. When in a bad cycle over a month; daily / weekly self assessment is skewed badly downward or data non-existent. That in itself can be useful if you identify that through lack of data capture there is a problem - but - to notice that trend you need to be looking for or alerted to the gaps and trend.

The system I have been trying to develop the last few months is based on capturing data from my family, friends and some work colleagues. It is a passive system that only works by not looking directly at the responses but aggregating together the information and has a few alerts built in. So both of my sons and my wife have a little app on their phone which simply tags Dad in a good mood - bad mood - indifferent.

There is a twitter account which people post - you are in a bad mood - you are in a good mood - other comments ignored and deemed indifferent - only the computer reads them. I also have an app on my phone that randomly asks me how I feel - once or twice a week - random times of day - that is also the mechanism to alert me there may be a problem - an alert triggers the app to ask how I feel.

From these disparate external feeds - I have a little script that grabs the overall good / bad / indifferent and plots it. Family ratings get weighted higher. Lots of bad mood points in succesion kick in weighting - lots of good mood points in succession kick in weighting. If there is a downward trend an alert kicks in if there is an upward trend an alert kicks in. I do not see the individual data points or where they came from - just the aggregated trending. If people on my monitoring list dont stick in data - they get a little reminder asking for a quick checkpoint - general cadence at the moment is between 3-10 days from each of the observers.

Obviously the whole system relies on peoples willingness to help provide the data points - and it still has a lot of work to tweak the model - but I can tell you now that some of the bad mood trend alerts hit me at just the right time and helped me get back into mindfulness that day.

Im hoping to get the model working to a point where with sufficient external observers putting in the data points at a not too annoying (for them!) interval - the system can help alert you to take corrective action - or in extreme cases might alert your carer / doctor.

Teenage son hitting bad mood 43 times when I turned off the xbox did skew the data - but can work on that :slight_smile:

This is my second post, based on my discovery of this site and these groups. I was in the middle of trying to figure out a way to quantify my self for my own mood tracking. After reading many of the posts, I am wondering if the process of quantifying one’s self is actually a motivator, an inhibitor, or a distraction from working through situations and their effect on one’s mood. Seeing this site and mass of information is exciting and makes me ‘feel’ energized about the act of tracking myself and possibly getting some answers to questions about myself. However, I can foresee spending so much time on the process that I never get to the end goal, of really determining my mood rhythms and affects, and finding a way to work on improving my mood.

Is this a typical response to the concept of the quantified self? Am I just ‘feeling’ overwhelmed?

In reading these ‘older’ posts discussing indirect mood measures, I am wondering if the tracking of inability to determine one’s mood would be a good indirect measure. For someone that needs to track mood in a quantified manner, it would seem to be possible that there is an inability to determine one’s mood in general. i.e. I am having a bad day, I am angry and short-tempered, but I don’t really know why. The situation of being so disconnected from one’s psyche that you cannot determine your own mood could be a good indicator of a track-able factor.

In any case, it is important to track the trends in moods, but the causes of these changes may be the very thing that should be tracked… such as blood sugar level, sleep/rest rate, weather, social circumstance, etc.

And with regards to having other people provide data points, I suppose if learning to communicate these data points is not an option, then having an app is a good alternative.


Hi, welcome, and I hope the forum proves useful! My one suggestion for starting out a project like this is to just go ahead and begin withs something really simple. For instance, we had a great talk from somebody who just gave themselves a single score on a scale of 1-10 first thing in the morning. He just asked himself the question: How do I feel today? There was actually a lot of good insight available to him in seeing how this score changed over time.

That’s one example of how it doesn’t have to be too hard to get started.