Is time-shrinking a new illusion ?

[Yoshitaka Nakajima <nakajima@xxxxxxxxxxxxxxx>]

Dear Colleagues,
     I learned that some of my colleagues are interested in whether my
coworkers and I are prepared to take a reaction against the criticism
(in P&P) on our articles, where we reported a 'new' phenomenon which we
call 'time-shrinking'.  Suppose that two neighboring empty durations
marked by three sound bursts have physical durations of 50 ms and 120 ms
(in this order).  The second duration is extremely underestimated (The
PSE is about 70 ms, showing an underestimation of 40 %).  The point of
the criticism is that this phenomenon is not novel because this is
'just' a sort of positive time-order error.
     We reacted, actually, and our reply was not accepted, partly due
to my poor language maybe.  We decided that the best thing we can do
now is to publish the data we have gathered as quickly as possible in
order to show the curious nature of this phenomenon.  Before doing this
tough job, I would like to summarize some points of our (almost
rejected) reply:

1) No previous data or theories could predict such a systematic, stable
   and big effect of so simple a temporal context on duration
   perception.  Tendencies found in studies of time-order errors or
   related materials are rather inconsistent.  If an unpredictably
   clear tendency is established in this almost chaotic situation, it
   should at least be recognized as a step to a new finding.

2) Even when a mathematical equation has proved useful to describe
   experimental results obtained in a particular paradigm, we often
   have to choose several parameter values for each experimental
   condition which fit the empirical data, and people often forget
   that this procedure also makes an 'ad hoc' explanation.  We do not
   think time-shrinking could be derived from previous theories.
   (This is the point where we are having trouble.  It seems
   logically impossible to prove that these 'methematical' models
   'cannot' predict the phenomenon.)

3) Time-shrinking disappears suddenly when the difference between the
   first and the second duration exeeds about 100 ms, and this
   tendency is constant even when the first duration is varied between
   40 ms and 280 ms.  This clear, but peculiar, tendency is difficult
   to derive from conventional studies of time-order errors.

4) In most studies of time-order errors, inter-duration intervals above
   500 ms are used, and it seems difficult to relate the experimental
   data obtained in such conditions directly with the data of our
   previous experiments, where the inter-duration intervals were 0 ms.
   We have a set of unpublished data (which we presented in Bourges,
   France, in 1992) showing that the underestimation of the second
   duration suddenly increases when the inter-duration interval becomes
   shorter than a certain value below 400 ms.  The data suggest that
   the positive time-order error obtained in a conventional paradigm
   and our time-shrinking have different mechanisms.

     I would appreciate any opinions either to me directly or to this
network.

                                       Best regards,
                                       Yoshitaka Nakajima
                                       nakajima@kyushu-id.ac.jp