從小鼠的前肢抓取任務探討過去任務結果如何影響大腦感覺運動皮質中內在狀態的表現與行為的變異

Abstract

內在狀態是指由生理需求、情緒與動機等內在變因所構成的整體狀態。一般認為，動物的行為會受到內在狀態的影響。然而，這些狀態如何在連續行為之間維持與演變，以及過去行為結果如何在下一次任務開始前表徵於感覺運動皮質，仍然缺乏充分的了解。本研究提出假說：過去的行為結果會影響動物的內在狀態，進而調控整個感覺運動皮質網路的基本神經動態，並影響後續行為表現。 為驗證此假設，我們訓練小鼠執行目標導向的抓取任務，並同步記錄次級運動皮質(M2)，初級運動皮質(M1)，與初級感覺皮質(S1)的神經活動。我們首先分析了由任務指示之行為與非任務指示之自發性動作的變異性，以探討內在狀態在行為層級的表現。結果顯示，多種動作的變異性會受到先前行為結果的影響，並且與當前任務之準備狀態相關，顯示內在狀態的變化能夠直接反映於行為的變異性。 在神經層級，過去行為結果與持續且廣泛分佈於整個感覺運動皮質的群體活動狀態相關。此狀態在下一次任務開始前即可被發現，並隨著任務進程持續的演變。在任務開始前，神經準備活動同時包含了過去行為結果，以及對當前任務之結果具有預測性的資訊。因此，此時的神經群體活動狀態提供了過去行為結果與後續行為變異之間之間可能的神經連接。 在單一神經元層級，我們發現僅有少部分神經元呈現簡單且具有選擇性的結果狀態相關反應，多數具有結果相關資訊之神經元呈現較為異質且複雜的反應模式。然而，僅利用這些神經元，即可有效解碼出當前與先前試驗的結果，顯示結果相關資訊並非由少數專一神經元單獨承載，而可能是以分散且混合的神經訊號共同承載於感覺運動皮質中。 綜合而言，本研究顯示，感覺運動皮質不僅編碼正在進行的動作，也持續承載一種與結果歷史相關的神經群體狀態，將近期經驗與後續動作的準備及變異性連結起來。此持續性的群體活動可能構成內部狀態影響後續行為的重要神經基礎。

Internal state refers to a combination of internal variables, such as physiological needs, emotions, and motivation. Animal behavior is believed to be shaped by internal states. Yet how such states are maintained from one action to the next—and how recent action outcomes are represented in the sensorimotor cortex before the subsequent action begins—remains poorly understood. Here, we hypothesized that outcome history shapes internal state, which in turn modulates baseline neural dynamics across the sensorimotor hierarchy and biases subsequent actions. To test this, we monitored neural activity across the secondary motor (M2), primary motor (M1), and primary somatosensory (S1) cortices during a goal-directed reach-and-grasp task. We first investigated behavioral manifestation of internal states by analyzing the variability of both task-instructed and task-uninstructed movements. Our results demonstrate that the variability of these movements is systematically modulated by the combination of prior outcomes, defined as successful or failed food grabbing and consumptions, and current-trial preparatory state, suggesting that internal state fluctuations are directly reflected in behavior. Critically, recent action outcomes were associated with a persistent and distributed sensorimotor cortical population state that was detectable before initiation of the next movement and evolved throughout the trial. Preparatory population activity contained information about both the preceding outcome and the upcoming trial outcome, thereby providing a neural link between outcome history and subsequent behavioral variability At the single-neuron level, relatively few units exhibited selective outcome-dependent responses, whereas many showed heterogeneous or mixed outcome-related activity. Nevertheless, population decoding based on these mixed-selectivity units recovered both preceding and current trial outcomes, demonstrating that outcome-related information is distributed across the sensorimotor cortical population. Together, these findings reveal that the sensorimotor cortex does not merely encode ongoing movement but continuously carries an outcome-history-dependent population state that links recent experience to the organization and variability of future actions. This persistent population activity may therefore constitute a neural substrate through which internal states influence subsequent behavior.