"Long ago", a psychologist did the following experiment on children of different ages: he'd lay out one line of M&Ms with broadly uniform spacing, and another line, longer, with the same spacing, and ask the child "which line has more M&Ms?" After they correctly indicated that the longer line had more M&Ms, he'd adjust the spacing between the M&Ms in the shorter line such that it was now longer. Then he'd ask again, "which line has more M&Ms?" He found that, below a certain age (I hope this finding was approximate), children would indicate the longer (and sparser) line as now having more M&Ms, and above the threshold age they would point to the correct, denser line.
His published papers were viewed as establishing a major threshold in child development, where children started to be capable of assessing reality independently. "Not so long ago", someone decided to replicate this landmark finding.
The twist was, in the replication, instead of asking the children "which line has more M&Ms now?", they said "OK, now you can take one of the lines and eat all of the M&Ms that were in it".
And the threshold effect disappeared completely. Children of all ages chose the line with numerically more M&Ms, regardless of linear extent.
When the problem is your experimental design, you're not warranted in drawing conclusions. You've already demonstrated that you're not qualified to speculate on them.
> that still doesn't explain this problem with bad communication.
> There are two statements, P and Q, and you have a question of what's more probable: statement P or the conjunction P&Q. The experimenter (communicating badly) makes you believe that Q has a very high probability. However, it is still logically impossible for P&Q to be more probable than P.
Your objection only makes sense if (1) human-to-human communication consists of messages that precisely specify their intended meaning, and (2) people believe that (1) is true. (1) and (2) are both false.
"Long ago", a psychologist did the following experiment on children of different ages: he'd lay out one line of M&Ms with broadly uniform spacing, and another line, longer, with the same spacing, and ask the child "which line has more M&Ms?" After they correctly indicated that the longer line had more M&Ms, he'd adjust the spacing between the M&Ms in the shorter line such that it was now longer. Then he'd ask again, "which line has more M&Ms?" He found that, below a certain age (I hope this finding was approximate), children would indicate the longer (and sparser) line as now having more M&Ms, and above the threshold age they would point to the correct, denser line.
His published papers were viewed as establishing a major threshold in child development, where children started to be capable of assessing reality independently. "Not so long ago", someone decided to replicate this landmark finding.
The twist was, in the replication, instead of asking the children "which line has more M&Ms now?", they said "OK, now you can take one of the lines and eat all of the M&Ms that were in it".
And the threshold effect disappeared completely. Children of all ages chose the line with numerically more M&Ms, regardless of linear extent.
When the problem is your experimental design, you're not warranted in drawing conclusions. You've already demonstrated that you're not qualified to speculate on them.
> that still doesn't explain this problem with bad communication.
> There are two statements, P and Q, and you have a question of what's more probable: statement P or the conjunction P&Q. The experimenter (communicating badly) makes you believe that Q has a very high probability. However, it is still logically impossible for P&Q to be more probable than P.
Your objection only makes sense if (1) human-to-human communication consists of messages that precisely specify their intended meaning, and (2) people believe that (1) is true. (1) and (2) are both false.