Some therapies assign tasks between sessions and others do not, and the difference follows directly from what each approach believes produces change.
The session is a small fraction of the week
An hour a week is a negligible portion of a person's time, and it occurs in a setting unlike any in which the difficulty actually arises.
Approaches that treat change as a matter of new learning therefore have to extend beyond the hour, since learning requires practice in the relevant conditions.
Between-session tasks are the mechanism for that, which is why they are integral to some approaches rather than being an optional extra.
Learning is context-dependent
A skill acquired in one setting transfers imperfectly to others, which is a general finding in learning research rather than a peculiarity of therapy.
Someone who can describe a technique fluently in a consulting room may be unable to apply it in the situation that provokes the difficulty.
Practising in the real setting is what closes that gap, and it cannot be done inside the session by definition.
Some tasks are the treatment rather than practice
In treatments for anxiety, graded exposure to avoided situations is the active component, and most of it necessarily takes place outside sessions.
Behavioural activation for depression similarly works through scheduled activity across the week, with the session used to plan and review.
In these cases the between-session work is not reinforcement of therapy but the therapy itself, with the appointment functioning as supervision.
Records change what is available to work with
Memory for internal states is reconstructive and reliably distorted by current mood, so recollection at the next appointment is a poor record.
Logging thoughts, activities or sleep close to the time produces information neither party would otherwise have, including patterns invisible at weekly resolution.
Which is why monitoring is frequently the first task assigned, before any change is attempted at all.
Non-completion is treated as material
Tasks are commonly not done, and evidence links completion with better outcomes, which makes non-completion worth examining rather than reprimanding.
The reasons are usually informative, whether the task was too large, poorly understood, or interfering with something the person had not mentioned.
Approaches that do not assign tasks are not thereby less demanding, since their work occurs through the relationship and within the session, which is a different mechanism rather than a lighter one.