Direct and indirect observation
Observation may be direct, i.e., the researcher is the observer, recording what he or she is watching, or
Observation may be indirect, i.e., the researcher must rely on the reported observations (including self-observations) of others.
However, in social research (and, hence, in information research), observation may be direct or indirect. That is, the researcher him- or herself, may watch what is happening, or may rely on the reported observations of others.
For example, if you are interested in how people use Web search-engines, you could sit beside them during a surfing session and watch what happens, recording the different terms used for the subject and the stages the person went through. You could also log some of this information automatically - that is, the machine makes the observations for you - or you could videotape what was appearing on the screen, while you tape-recorded the person talking aloud about their activity. This would be direct observation.
On the other hand, you could conduct interviews with people about how they use search-engines: they would then have to recall what they did and report it to you. This assumes that people have an ability to recall earlier behaviour accurately - which may not always be true. If you then proceed to ask them about their opinions of, or attitudes towards Web search-engines, you are asking them to observe (probably for the first time) their mental states on these issues - you are asking them to make self-observations.