Question regarding GROUP BY

Hi,

I've got 3 tables:
objects (object_id integer primary ...)
projects (project_id integer primary ...)

in the 3rt table I store an m:n relation
obj_2_proj (object_fk, project_fk, access_ts timestamp, primary
key (object_fk, project_fk))

Now I need to know the projekt and access_ts of all those objekt_fk with
the highest access_ts.
This highest access_ts might be NULL.
I tried:

SELECT object_fk, project_fk, max(access_ts)
FROM obj_2_proj
GOUP BY object_fk;

Postgres doesnt like this and complains, I had to include project_fk in
the GROUP BY but if I do this I get every line out of this table since
(object_fk, project_fk) is the primary key.

What to do?

Regards
Andreas