|Abstract: ||視覺訊息整合與連結的議題 (binding problem) 旨在探討多項不同的視覺資訊是如何被平行的整合為一個完整、統一的心智表徵：例如一個完整的物件、知覺的連貫性、甚或是一個完整的情節記憶。除了知覺層次的連結外，視覺資訊的連結是如何被持續的維持，而能夠成為短期或是長期記憶同樣是重要的研究課題。本研究聚焦於物體的顏色與形狀的特徵連結，以及該連結在視覺工作記憶背後的神經震盪 (oscillation) 機制。|
儘管特徵連結現象在我們的日常生活經驗中看似不費吹灰之力，先前研究卻顯示處理多特徵物體的過程不論在行為或是神經生理的層次都並非沒有代價。最近的一項腦造影研究顯示當多特徵物體消失在我們的視野後，維持這樣的特徵連結可能需要左側頂葉與顳葉腦區的參與。該研究結果與許多先前指出頂葉似乎在視覺訊息整合，以及整合之後如何維持在工作記憶之中，扮演關鍵角色的觀點相符。此外，數個電生理的研究顯示伽馬頻段的神經震盪 (gamma frequency oscillation) 與工作記憶的負擔以及特徵連結、連貫性的視覺感知皆有密切關係。這引發了一項假說：特徵連結視覺工作記憶，可能是藉著顳葉與頂葉以伽馬頻段震盪進行訊息溝通與傳送而產生。頂葉在這個心智過程中可能扮演著類似「黏著劑」的角色，負責在景象消失之後仍持續的將構成物體的各個特徵連結在一起。然而，這項假說的因果關係並無法僅藉著相關性的研究工具回答。因而，本研究採用伽馬頻段 (gamma frequency)的跨顱交流電刺激技術 (tACS, transcranial alternating current stimulation)，同時調節左側顳葉與頂葉腦區的神經震盪模式，藉此釐清這兩個腦區的神經震盪模式與特徵連結視覺工作記憶 (binding-VWM) 的因果關係。
實驗一的結果證實了先前研究的發現，相較單一特徵，多特徵連結的視覺工作記憶是較為困難的。本研究的實驗二觀察到了一個顯著的作業變項 (僅需記住形狀 vs. 須記住形狀與顏色的連結) 與電刺激變項 (異相位的伽瑪跨顱交流電刺激 vs. 假性刺激) 的交互作用。這個交互作用來自於跨顱交流電刺激能夠專一性的提升受試者在進行特徵連結工作記憶的表現。值得注意的是，該電刺激在行為上的調節效果橫跨了電刺激進行的時段以及電刺激關閉以後的實驗階段，顯示該刺激效果能夠在結束刺激後維持至少一段時間。實驗二的發現支持了特徵連結視覺工作記憶可能是藉著顳葉與頂葉以伽馬頻段震盪，進行訊息溝通與傳送而產生的假說。此外，本研究的數據也顯示，該電刺激的調節效果與原本受試者間既有的個體差異存在著交互作用。這項交互作用來自於本實驗所採用的異相位伽瑪跨顱交流電刺激僅能提升低表現組的工作記憶，卻無法再進一步幫助高表現組。為了進一步確立該電刺激參數在實驗二效果的專一性，本研究於實驗三採取和實驗二完全一樣的交流電頻段並施打在一樣的兩個腦區，但使兩個腦區接受同相位的電刺激。實驗三的結果顯示，不論是單一特徵的作業或是特徵連結的作業，皆沒有發現任何電刺激在高表現組或低表現組有任何的效果。這些結果顯示實驗二的伽瑪跨顱交流電刺激效果具有一定程度的專一性。
;The binding problem refers to the question of how multiple and different kinds of distributed information can be integrated in parallel to a unitary representation, such as an intact object, coherent perception, or even a memory episode. Beyond perception, the binding problem persists when one has to maintain such coherent representation of the world in short-term or long-term memory. The present study focuses on the neural oscillatory mechanism behind the binding processes that form and maintain color-shaped objects, which we refer to as Feature-Binding Visual Working Memory (Binding-VWM).
Although feature binding seems natural and automatic from our daily experiences, several previous studies have revealed that the processing of multi-featured object is not cost-free, both at the behavioral and neurophysiological level. Recent neuroimaging research has demonstrated that both the left temporal regions and the parietal cortex are involved specifically during the binding maintenance period. This is in line with several previous studies suggesting that the parietal region seems to play a vital role in visual grouping and binding-VWM. In addition, several electrophysiological studies have provided evidence showing that gamma frequency oscillation is related to WM load and is also relevant to feature binding and perception of coherent visual patterns. This motivates the hypothesis that feature binding in VWM may be mediated by the communication between temporal and parietal regions via gamma oscillations. Parietal regions here may provide the ‘glue’ that continuously maintain these bound features together even when objects are no longer in view. Causal evidence of how these brain regions operate together in storing multiple features of an object in VWM, however, has not been established with the use of correlational approaches. To this end, here we adopt gamma-frequency transcranial alternating current stimulation (tACS) to modulate the oscillatory signals between the left temporal and parietal cortices simultaneously to test the causality between their oscillatory pattern and binding-VWM performance.
In Experiment 1, the results confirmed previous research indicating the demanding nature of binding in VWM. In Experiment 2 there was a significant interaction between task condition (shape only vs. binding) and stimulation condition (out-phase gamma stimulation vs. sham), which was driven by the specific improvement via tACS in the binding condition but not in the sham condition. It is worth noting that the modulation effect was consistent across the on-line and off-line periods, suggesting that a carryover tACS effect was present even when stimulation stopped. The findings support the hypothesis that binding-VWM is established by the crosstalk between left parietal and temporal regions, in which gamma oscillation is a critical component. Moreover, our data demonstrate that the d’ profile of this modulation is not homogenous, but interacts with preexisting individual differences, where out-phase gamma tACS improved only low-performers’ memory. In Experiment 3 we applied in-phase gamma tACS over the same brain regions as Experiment2, but found no effect of tACS in both the single-feature and binding condition across low- and high-performers. These results suggest that the modulation effect from Experiment 2 is not a general effect of gamma tACS.
Taken together, our study showed that binding-VWM is demanding, but the low-performers can benefit significantly from out-phase gamma tACS over the left temporal and parietal regions, suggesting a possible mechanism of information transmission between these regions. It is likely that such neural interaction is what enables us to perceive and remember both the individual tree and the entire forest.