Thanks to the comments on this site to my question and because of continued reading of Kant’s Critique of Pure Reaons (CPR) I would like to give an answer to my own question.
Why do we need freedom as a transcendental idea – according to Kant?
Kant accepts the thesis of the third antinomy. It states:
Causality according to the law of nature is not the only causality from which all the appearances of the world can be derived. In order to account for theses appearances, it is necessary also to admit another causality, that of freedom. (CPR, B472)
In the proof of this thesis Kant argues: The law of causality which governs experience establishes an infinite chain of causes. But without a first cause each cause of this chain is up in the air and the whole chain does not constitute anything. As a consequence, one has to assume as hypothesis a first cause. It starts the whole chain of causes. The way this first cause acts, is named transcendental freedom (B474).
My short answer: Kant needs the hypothesis of transcencental freedom to accept the law of causality as valid for our experience.
Why does freedom in its practical sense explain our experience – according to Kant?
Freedom, in its practical sense, is the independence of our will from coercion through impulses of sensibility. (CPR B561)
According to Kant, freedom in its practical sense is established by the fact that we can start actions on the base of rational decisions. Also “independent of the necessitation through sensible impulses” (B561) - even against them, I would add. The rational decision initiates a new chain of causality.
The basis of this argumentation is a certain view on human rationality and decision making.
Eventually, I would like to criticize both claims of Kant.
First, Kant’s scope of thinking about causality is too narrow. The only type of causation Kant considers are monocausal linear chains. He does not take into account that a certain effect may presuppose more than one cause. Neither does Kant consider the distinction between necessary and sufficient causes. About which type of causes does Kant speak? Most of all, Kant does not consider cyclic causality in the form of feed-back chains. One can discuss, whether the hypercycle of Manfred Eigen, a model of the interaction of nucleic acids (carrier of information) and amino acids (protein), exemplifies a chain of causality without a first cause.
Of course, one can apply Kant’s argument with linear chains of causality to the standard model of cosmology. Here we must admit: We do not know how the concept of causality applies when extrapolated into the far past. Using this question, one can exemplify the different methods of metaphysics and physics: The former makes the hypothesis of a first cause, the latter leaves the question open for further scientific progress.
Secondly, Kant’s scope of rational decision making is too narrow. Kant considers man primarily as a rational being. For Kant rational mental processes are conscious processes. But most processes in our brain start as unconscious process, not reachable for our introspection.
Up to now we do not have a scientific model about how a decision derives from former experiences, about their evalution and about the unconscious alignment of conflicting alternatives of action. But such models are the subject of current neuroscience, working on a determistic basis. I consider these mechanisms an
“ignoramus” but not an “ignorabimus”.