I have been reading http://psych.stanford.edu/~jlm/pdfs/Norman76.pdf, which on pp34-36 present an excerpt from a book by Neville Moray.

Tone bursts of 3000 Hz were delivered to one ear and tone bursts of 2111 Hz to the other ear of observers. The onset and offset of the tone bursts were exactly simultaneous on the two ears, and controlled through a two-channel electronic switch."

Subjects were required to press a left hand button when they heard an increment in their left ear, a right hand button when they heard an increment in their right ear, and a central button when they heard an increment in both ears. Four conditions of presentation were used. These were Single Mode (in which only one ear was used, the other being silent throughout; Select Mode (in which both ears received tone bursts, but the listener was to ignore one ear completely and only respond to the other); XOR Mode (in which both ears received tone bursts, but targets never occurred simultaneously); and IOR Mode (in which targets could occur either on the left ear or on the right ear, or simultaneously on both, the last condition being a logical AND Mode). In addition a control was run using IOR presentation Mode but requiring the listener to ignore thesingle targets and respond only to the simultaneous pairs with the AND response.

It is rather curious that so little attention has been paid to the AND mode of presentation (simultaneous targets), Most of the experiments have been either Select Mode in which the listener accepts one message and rejects the other or, more rarely, XOR in which responses are made to targets in two messages, but targets are never simultaneous (Treisman and Geffen, 1967, for example), Indeed, while Deutsch. and Deutsch's model clearly predicts that the AND condition will show severe interference between the messages, Treisman's model makes no explicit prediction about AND signals, Treisman says that in the XOR condition and the Select Mode the listener can attenuate an unwanted message (hut there is no indication how rapidly such attenuation could be redirected in the XOR Mode). However, for the AND case, it is not clear that a listener should attenuate one and reject the other rather than rejecting both, or for that matter accepting-both, since the theory only says that the Select Mode can result in the attenuation of one message, not that in all competitive situations attenuation of all except one message must occur,

The Table [2.1] shows that Treisman's model will not fit the facts. Notice the last three rows of data, AND is the percentage of ANDs (simultaneous targets) which were correctly detected in the lOR presentation mode. AND'" is where IOR presentation was used with AND response only (other targets being ignored) , Here again responses to ANDs are extremely poor ... NAND is the percentage of single responses made either to a left target, or a right target, or to what was really an AND.

Now, if Treisman's model says that an: observer sharing his attention equally between two messages attenuates neither, AND should equal Singles. They do not. If he attenuates both messages ·when sharing equally between the two, NAND responses should be no better than AND responses, but they are. If the listener attenuates one channel and listens' only to the other, given that attention wandered from side to side the' observed results could be obtained; on the basis of the shadowing work Treisman maintains that information about crude physical characteristics is available from both messages even in Select Mode. But the data on AND responses shows that this is not true.

Treisman's model does not then fit the data, and needs more elaboration to state explicitly what the predictions are from it about performance in the AND Mode.

Table 2.1

As I see it Moray is arguing that since attention is being split evenly between ears, when targets occur simultaneously the Attenuation model says both the left target and right target should be attenuated equally. So then when they get through the filter and are analyzed for meaning they should either both be ignored (if they've been attenuated too much) or else both be noticed (if they haven't been attenuated too much). Since the attenuation would have been of the same magnitude if there was just a target in one ear, there's no reason for simultaneous targets to be harder than single targets. But there is, and so there's something wrong with Treisman's model.

Is that a correct understanding of Moray's argument?

  • $\begingroup$ Perhaps my confusion stems from a misunderstanding of this statement. "NAND is the percentage of single responses made either to a left target, or a right target, or to what was really an AND." What's a 'single response'? $\endgroup$ Oct 17 '13 at 10:21
  • $\begingroup$ Usually I understand "single response item" to mean a question that can have only one answer ("How old are you?"), as opposed to questions that can have more than one answer ("Which of the following books have you read?"). $\endgroup$
    – user3116
    Oct 17 '13 at 11:51

Neville Moray has given permission for his answer (provided via personal correspondence) to be quoted here:

"Treisman said that physical characteristics such as loudness or pitch were not attenuated by selective attention, because although you can't hear semantic information from rejected messages in shadowing experiments, you can hear changes in pitch, or tell a pure tone from speech. So according to her you should be able to listen to two different tones simultaneously. But my experiment showed that you can't."


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.