You should pay attention to key mental events, on a regular and frequent basis, because important thoughts can happen very briefly or very occasionally and you need to catch them.
You may find your understanding of this post significantly improved if you read the third story from Seven Shiny Stories.
Luminosity is hard and you are complicated. You can't meditate on yourself for ten minutes over a smoothie and then announce your self-transparency. You have to keep working at it over a long period of time, not least because some effects don't work over the short term. If your affect varies with the seasons, or with major life events, then you'll need to keep up the first phase of work through a full year or a major life event, and it turns out those don't happen every alternate Thursday. Additionally, you can't cobble together the best quality models from snippets of introspection that are each five seconds long; extended strings of cognition are important, too, and can take quite a long time to unravel fully.
Sadly, looking at what you are thinking inevitably changes it. With enough introspection, this wouldn't influence your accuracy about your overall self - there's no reason in principle why you couldn't spend all your waking hours noting your own thoughts and forming meta-thoughts in real time - but practically speaking that's not going to happen. Therefore, some of your data will have to come from memory. To minimize the error introduction that comes of retrieving things from storage, it's best to arrange to reflect on very recent thoughts. It may be worth your while to set up an external reminder system to periodically prompt you to look inward, both in the moment and retrospectively over the last brief segment of time. This can be a specifically purposed system (i.e. set a timer to go off every half hour or so), or you can tie it to convenient promptings from the world as-is, like being asked "What's up?" or "Penny for your thoughts".
When you introspect, there is a lot to keep track of. For instance, consider the following:
- What were you thinking about? (This could be more than one thing. You are a massively parallel system.) Was it a concept, image, sensation, desire, belief, person, object, word, place, emotion, plan, memory...?
- How tightly were you focused on it? (Is the topic itself narrow or disparate?) What other items (sensory, cognitive, emotional) seemed to intrude on your concentration, if any, and how did you react to this incursion?
- How did you feel about the subject of the thought? This includes not only emotional reactions like "this is depressing" or "yay!", but also what you felt inclined to do about the topic (if anything), and how important or interesting your thought seemed.
- How does thinking, in general, feel to you? (I conducted an informal survey of this and got no two answers the same. Anecdotally, it may be rather key to determining how you are different from others, and so in refining your model of yourself relative to the fairly generic priors we're starting with.) Coming up with a good way to conceptualize your style of thinking can help you interpret introspective data, although be sure to abandon a metaphor that looks about to snap. You might have different answers when you're "actively" thinking something through - i.e. when novel information is generated in your mind - and when you're thinking "passively", as when you read or listen to some information and absorb its content as it comes.
- What memories did the thought dredge up, if any - parallel situations from the past, apparently unrelated anecdotes that floated by for no reason, events where you learned concepts key to the topic of your thought? Did the thought generate anticipations for the future - a plan, a fear, a hope, an expectation, a worry?
- What sensory input were you receiving at the time? Include not only sight, sound, smell, touch, and taste, but also things like temperature, proprioception, and internal symptoms like hunger or nausea. Can you determine how, if at all, that interacted with the thought?
You cannot have too much data. (You probably can have too much data in one situation relative to how much you have in another, though - that'll overbalance your models - so make a concerted effort to diversify your times and situations for introspection.) When you acquire the data, correlate it to learn more about what might bring various aspects of your thought into being.