Slashdot Mirror
A Dolphin By Any Other Name
SloppyElvis writes "CNN is reporting that scientists have proven that Dolphins can communicate with each other by name. From the article: 'researchers synthesized signature whistles with the caller's voice features removed and played them to dolphins through an underwater speaker' to which the mammals responded. This form of identification in language was previously only known to exist in the human world." Thankfully they still haven't evolved opposable thumbs.
But that's just because they're in the water.
You better watch out, there may be dogs about . .
On a related note, what the heck does "clinically proven" mean?
It usually means that a randomized clinical trial was performed and the null hypothesis was rejected. If you haven't had any statistics yet, that means that you hypothesize two treatments have the same effect and run a trial. If the results observed would be highly improbable under the equal effect hypothesis (and this probability is very clearly defined), then you have 'clinically proven' treatment A superior, for instance. Statistical hypothesis testing is basically applying the same principle as a reductio ad absurdum from logic. In general, you usually don't 'prove' things in science like you do in math, but that's another topic.
...it can be proven beyond reasonable doubt.
"reasonable doubt" is a legal term, not a scientific one.
:wq
Although it is still improper to state 'proven'.
.05). American Psychologist, 49, 997-1003 [Cohen is recognized as a fairly statistically savvy person in psychology circles]), which essentially make it impossible to state that you have 'proven' a theory or even a specific hypothesis. I won't go into the details, but essentially most hypothesis testing accepts p .05 as the necessary alpha level for rejection of the null hypothesis, which is essentially stating "I am going to say that this hypothesis is correct, but there is a 5% chance that my results are due to random error and variance in my sample, and not really a repeatable phenomenom."
.05 (5%), but are occasionally set at .1, .01, or even .001. The .05 tradition is strong, however, and hails back to the days when statistics were being used in agriculture to develop varying pesticides and the like. Realistically, whatever the researcher is willing to live with is generally acceptable, but more than a few journals get a little tetchy about any probability higher than .05 (Journal of Applied Psychology being one, IIRC). The peer review process keeps this quite sane, but here's a sobering thought:
Generally it is considered bad form and logically incorrect. Statistical hypothesis testing has its own pitfalls (see Cohen, J. (1994). The earth is round (p
It should be noted that the parent is correct, but that there are some alternative methods available (although these are rarely used), such as Bayesian Statistics and others. Bayesian Statistics is new to psychology, and I really don't have the background to evaluate it (read: I've never studied it), but I've heard good things about the concept.
FWIW, the probabilities stated by the parent are generally set at
In my master's thesis, I ran about 50 correlations on about 100 participants, each providing about 20 data points. I only had 5 'significant' (at p.05) correlations). THis means that half of my correlations could considered likely to be just chance--random variance. But, because of the nasty nature of statistics, I had no way of knowing which ones were due to chance and which ones were not--thus the whole study was considered as being unable to reject the null hypothesis (no relationships). Not particularly exciting, but informative nonetheless. Unfortunately, I also don't know what the probability of falsely accepting the null was (power), which means that I can't be certain that these are correct either.
"We don't know what we are doing, but we are doing it very carefully,..." Wherry, R.J. Personnel Psychology (1995)
While whale-watching in the North Pacific ocean around San Juan Island, the whale expert explained how whales make a unique sound before and after their other phrases, and that these sounds are often accompanied by a reply for a different whale. The unique sounds were most often unique to the whale that responded. As such, experts believed these to be used like names.
Such a conversation would go something like this:
There's evidence that dolphins ARE actually self-aware; this is easily proven by putting a mirror in front of them. If they attack the mirror or run away from it, they think it's another animal. If they stare at it, and then turn about and inspect parts of their bodies not normally visible to them, they're aware that the thing in the mirror is them.