Quantcast

How can learning initiate? -Comments on Amitay’s work

Amitay at MRC (now @ UK London) presented his work about auditory PL in latest Nature Neuroscience, a well-known neuroscience journal.

In that paper (Amitay,Irwin & Moore, 2006), they tested the discriminability of tone frequency through a short-term probe procedure (30 trials) in 120 subjects, allocated in 10 groups equally, following 10 varied training procedures (100 trials/block*8 block). There are three probe tests and eight training sessions (see following Figure 1a). The task is a 3-interval tone frequency discrimination task. The reference frequency is 1 KHz if unstated.

They found that three adaptive training paradigms, manipulating performance at 50% (difficult), 75% (medium) and 95% (easy), significantly improved subjects’ discriminablity (Figure b and c). Moreover, the learning effects are comparable. Thus,varing the level of difficulty has no effect of learning.

Is there differences in learning between varying and unvarying training stimuli? They tested the possibility by training subjects in three fixed tone frequency differences, 400 Hz (very easy,ceiling), 7 Hz (difficult, near threshold after 1000 trials’ training) and 0 Hz (change performance, the reference and test tone are actually the same). They found there is significant learning that matched the magnitudes of varying training paradigms. Here, what surprised me most is the significant learning under 0 Hz difference — just repetitive presenting references. It can be simply explained as a stimulus-driven/bottom-up learning mechanism. Learning under 7 Hz difference is significantly robust than under 400 Hz difference, which they attributed to difference of tone exposures.

They also tested whether learning under other modalities can enhance auditory? During “NONE” condition under which subjects only read books, there is no learning. Under Tetris– a visuospatial game, learning is obvious. If subjects play Tetris while listen to the playback of stumuli used for a subject that trained with varying paradigm (75%), they learned significantly and even surprisingly, learned more on the Tetris game. The authors did not give an explanation. To me, it may reflect difference in arousal for experimenter asked subjects to ignore the interfering tone and concentrate on the Tetris game — the subjects might be more attented to the game and thus more arousal during the training/test periods.

In addiiton, if trained with varying paradigm that trackes 75% correct to discriminate a stimulus from a 4 KHz tone , the learning effects transferred significantly to 1 kHz discrimination task. Off-frequency (all three 4 kHz frequency, zero difference) training can not triger learning.

They thus stated that three processes are necessary for optimal perceptual learning: sensitization through exposure to the stimulus, modality- and dimension-specific attention, and general arousal.

To me, there are several points that should be clarified/corrected:

1) the Tile “Discrimination learning induced by training with identical stimuli”. Actually, it reflected limited info. It is kind of showing-off. A good tile should be ^^^^^(everybody can provide results of his own).

2)I wanna know the detail of learning course. Is it largely individual? How many trials that will first make the learning asymptote? This seems a ‘too’ easy task that differs significantly from most visual learning. Anybody knows the condition in the field of auditory PL? So few trials are questionable.

3) In another Nat.Neurosci. paper, Amitay et al. addressed that most procedural learning will be perceptual per se. It actually mismatched with findings in this paper. They found in this paper that pure learning in Tetris game can induce significant learning. It should not if I am not wrong. According to his last NNS paper, what on the earth have learned are actually perceptual, i.e. is specific to the trained dimension and modality and thus cannot transfer to the tone frequency discrimination task. However, it happened. To me, it only results from repetitive probe trials. The result from the subjects that only read book seems to undermine my conjesture. However, the author must provide the between-group test. Anyway, if I am wrong, the effects observed in Tetris playing subjects cannot be attributed to repetitive probe and it reflects genuine generability, the results contradict with his previous paper on the other hand.

4) I wanna know how the authors probe the threshold?

5) Is there actually no difference between those groups, such as varying, invarying and Tetris groups? Is there difference in learning rate? The task is too easy to compare, I think.

6) ^^^^^^^^^^ (to be continued by me or YOU)

APPLAUDS & COMMENTS are both wel. :>




The material in this press release comes from the originating research organization. Content may be edited for style and length. Want more? Sign up for our daily email.