Limits...
The impact of early visual cortex transcranial magnetic stimulation on visual working memory precision and guess rate

View Article: PubMed Central - PubMed

ABSTRACT

Neuroimaging studies have demonstrated that activity patterns in early visual areas predict stimulus properties actively maintained in visual working memory. Yet, the mechanisms by which such information is represented remain largely unknown. In this study, observers remembered the orientations of 4 briefly presented gratings, one in each quadrant of the visual field. A 10Hz Transcranial Magnetic Stimulation (TMS) triplet was applied directly at stimulus offset, or midway through a 2-second delay, targeting early visual cortex corresponding retinotopically to a sample item in the lower hemifield. Memory for one of the four gratings was probed at random, and participants reported this orientation via method of adjustment. Recall errors were smaller when the visual field location targeted by TMS overlapped with that of the cued memory item, compared to errors for stimuli probed diagonally to TMS. This implied topographic storage of orientation information, and a memory-enhancing effect at the targeted location. Furthermore, early pulses impaired performance at all four locations, compared to late pulses. Next, response errors were fit empirically using a mixture model to characterize memory precision and guess rates. Memory was more precise for items proximal to the pulse location, irrespective of pulse timing. Guesses were more probable with early TMS pulses, regardless of stimulus location. Thus, while TMS administered at the offset of the stimulus array might disrupt early-phase consolidation in a non-topographic manner, TMS also boosts the precise representation of an item at its targeted retinotopic location, possibly by increasing attentional resources or by injecting a beneficial amount of noise.

No MeSH data available.


Cartoon summary of results.The left most panel depicts an example recall distribution. The frequency of recall errors (i.e. difference between response and target orientation) is indicated by the height of the grey bars. A mixture model fit of the recall errors is depicted in thick orange lines. Most responses are centered around 0°, and the precision of memory recall is captured by the mixture model SD parameter (smaller SD indicates more precision). Some proportion of responses appear to be random, indicated by the probability of guess-like responses (smaller p-Uniform indicates less guessing). The two rightmost panels provide a cartoon summary of our main results (exaggerated and simplified for illustrative purposes). First, the main effect of relative location is highlighted by increased memory precision (SD) at the location targeted by TMS (‘same’) compared to the control quadrant (‘diagonal’) for both early and late TMS pulses (compare SD at four visual field locations). Second, the main effect of pulse timing is shown by increased guessing when pulses were applied early compared to pulses applied late irrespective of visual field location (compare p-Uniform between middle and right most panels).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5383271&req=5

pone.0175230.g005: Cartoon summary of results.The left most panel depicts an example recall distribution. The frequency of recall errors (i.e. difference between response and target orientation) is indicated by the height of the grey bars. A mixture model fit of the recall errors is depicted in thick orange lines. Most responses are centered around 0°, and the precision of memory recall is captured by the mixture model SD parameter (smaller SD indicates more precision). Some proportion of responses appear to be random, indicated by the probability of guess-like responses (smaller p-Uniform indicates less guessing). The two rightmost panels provide a cartoon summary of our main results (exaggerated and simplified for illustrative purposes). First, the main effect of relative location is highlighted by increased memory precision (SD) at the location targeted by TMS (‘same’) compared to the control quadrant (‘diagonal’) for both early and late TMS pulses (compare SD at four visual field locations). Second, the main effect of pulse timing is shown by increased guessing when pulses were applied early compared to pulses applied late irrespective of visual field location (compare p-Uniform between middle and right most panels).

Mentions: While participants were remembering four orientations, 10Hz triple-pulse TMS was applied over early visual cortex retinotopically corresponding to the location of one of the to-be-remembered items. Orientation recall (i.e. the absolute error) differed between the four locations at which stimuli had been presented, likely due to better performance at the location targeted by TMS compared to the location diagonal to TMS. Additionally, recall was marginally worse for early (directly at stimulus offset) compared to late (midway through retention) pulses. Recall errors were fit with a mixture model to reveal relative contributions of changes in memory variability on the one hand, and the probability of guessing responses on the other: Retinotopically specific improvements proximal to the pulse were attributed to reduced response variability, implying that memory precision can be improved locally by means of TMS. Global impairments for early compared to late pulses were due to increased guess rates, implying non-retinotopic disturbances due to TMS at the tail end of encoding. A cartoon-summary of these findings is shown in Fig 5. None of these findings were observed with sham TMS, and it is unlikely that differences in performance at the four stimulus locations due to visual field anisotropies were driving the observed TMS effects.


The impact of early visual cortex transcranial magnetic stimulation on visual working memory precision and guess rate
Cartoon summary of results.The left most panel depicts an example recall distribution. The frequency of recall errors (i.e. difference between response and target orientation) is indicated by the height of the grey bars. A mixture model fit of the recall errors is depicted in thick orange lines. Most responses are centered around 0°, and the precision of memory recall is captured by the mixture model SD parameter (smaller SD indicates more precision). Some proportion of responses appear to be random, indicated by the probability of guess-like responses (smaller p-Uniform indicates less guessing). The two rightmost panels provide a cartoon summary of our main results (exaggerated and simplified for illustrative purposes). First, the main effect of relative location is highlighted by increased memory precision (SD) at the location targeted by TMS (‘same’) compared to the control quadrant (‘diagonal’) for both early and late TMS pulses (compare SD at four visual field locations). Second, the main effect of pulse timing is shown by increased guessing when pulses were applied early compared to pulses applied late irrespective of visual field location (compare p-Uniform between middle and right most panels).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5383271&req=5

pone.0175230.g005: Cartoon summary of results.The left most panel depicts an example recall distribution. The frequency of recall errors (i.e. difference between response and target orientation) is indicated by the height of the grey bars. A mixture model fit of the recall errors is depicted in thick orange lines. Most responses are centered around 0°, and the precision of memory recall is captured by the mixture model SD parameter (smaller SD indicates more precision). Some proportion of responses appear to be random, indicated by the probability of guess-like responses (smaller p-Uniform indicates less guessing). The two rightmost panels provide a cartoon summary of our main results (exaggerated and simplified for illustrative purposes). First, the main effect of relative location is highlighted by increased memory precision (SD) at the location targeted by TMS (‘same’) compared to the control quadrant (‘diagonal’) for both early and late TMS pulses (compare SD at four visual field locations). Second, the main effect of pulse timing is shown by increased guessing when pulses were applied early compared to pulses applied late irrespective of visual field location (compare p-Uniform between middle and right most panels).
Mentions: While participants were remembering four orientations, 10Hz triple-pulse TMS was applied over early visual cortex retinotopically corresponding to the location of one of the to-be-remembered items. Orientation recall (i.e. the absolute error) differed between the four locations at which stimuli had been presented, likely due to better performance at the location targeted by TMS compared to the location diagonal to TMS. Additionally, recall was marginally worse for early (directly at stimulus offset) compared to late (midway through retention) pulses. Recall errors were fit with a mixture model to reveal relative contributions of changes in memory variability on the one hand, and the probability of guessing responses on the other: Retinotopically specific improvements proximal to the pulse were attributed to reduced response variability, implying that memory precision can be improved locally by means of TMS. Global impairments for early compared to late pulses were due to increased guess rates, implying non-retinotopic disturbances due to TMS at the tail end of encoding. A cartoon-summary of these findings is shown in Fig 5. None of these findings were observed with sham TMS, and it is unlikely that differences in performance at the four stimulus locations due to visual field anisotropies were driving the observed TMS effects.

View Article: PubMed Central - PubMed

ABSTRACT

Neuroimaging studies have demonstrated that activity patterns in early visual areas predict stimulus properties actively maintained in visual working memory. Yet, the mechanisms by which such information is represented remain largely unknown. In this study, observers remembered the orientations of 4 briefly presented gratings, one in each quadrant of the visual field. A 10Hz Transcranial Magnetic Stimulation (TMS) triplet was applied directly at stimulus offset, or midway through a 2-second delay, targeting early visual cortex corresponding retinotopically to a sample item in the lower hemifield. Memory for one of the four gratings was probed at random, and participants reported this orientation via method of adjustment. Recall errors were smaller when the visual field location targeted by TMS overlapped with that of the cued memory item, compared to errors for stimuli probed diagonally to TMS. This implied topographic storage of orientation information, and a memory-enhancing effect at the targeted location. Furthermore, early pulses impaired performance at all four locations, compared to late pulses. Next, response errors were fit empirically using a mixture model to characterize memory precision and guess rates. Memory was more precise for items proximal to the pulse location, irrespective of pulse timing. Guesses were more probable with early TMS pulses, regardless of stimulus location. Thus, while TMS administered at the offset of the stimulus array might disrupt early-phase consolidation in a non-topographic manner, TMS also boosts the precise representation of an item at its targeted retinotopic location, possibly by increasing attentional resources or by injecting a beneficial amount of noise.

No MeSH data available.