| Decision making is a process by which animals make their choices during their living or surviving.In most circumstances,decision making involves the evaluation of the outcome of certain behavior.Psychologists have concluded that the key drive of human’s decision-making is emotion,which occasionally causes irrational and unbeneficial decisions.In recent researches,it has been demonstrated that the medial frontal cortex(mPFC),which located in the front of the frontal lobe,participates the process of decision making in two major ways.On one aspect,mPFC mediates animals’decision-making behaviors from an upper position through the bidirectional projections with most primary sensory cortexes and cortexes or subcortical nucleus involving more complex functions such as memory,cognition,emotion,reward et cetera.Meanwhile,mPFC handles the information of expectations,evaluations and feedbacks of outcomes during decision making,together with orbital prefrontal cortex(OFC),striatum and anterior cingulate cortex(ACC)functionally.So far,most researches concentrating on the mediating mechanisms of mPFC for decision-making behaviors are conducted on primates like human or monkeys.And the spatial resolution of commonly used imaging techniques including fMRI is not high enough for distinguish the firing patterns of mPFC neural populations.Therefore,which neuron group and how it contributes to the alteration of the animal’s decision is yet to be explored.Based on existing behavioral tasks,we designed a simplified Iowa gambling task protocol for mice.The tasks were completed in a manufactured automatic behavioral experiment device.We proved that mice were capable of discovering and executing the most beneficial strategy after certain number of trials.And this progress was made by constantly updating their behavioral patterns on the basis of reward or punishment feedbacks.Then we recorded the calcium signals of neural populations through GRIN lenses implanted at mice’s mPFC.Calcium activities were captured by a portable microscope mounted on the skull of mice simultaneously with mice’s IGT behavioral footages.Through integration with open-source algorithms,we developed a toolkit to finish the analysis and result visualization of calcium imaging data.We recorded 761,548 and 585 neurons in preference index testing phase,early IGT phase and late IGT phase.Results suggested that different neural populations were specifically related to individual behaviors in IGT.For instance,neurons responsive to trigger behaviors which initialized IGT trials and those responsive to reward obtaining didn’t overlap.They recruited 24%and 25%of total neurons separately.81%of trigger responsive neurons and 96%reward responsive neurons were selective on the locations where the mouse received a reward.During IGT phase,neurons,which were responsive to reward or empty reward obtaining,could be divided into several neural clusters according to their firing patterns.Approximately,there were three different patterns of neural activities:firing before a behavior,firing immediately at the onset of a behavior and delayed firing one to two seconds after a behavior.The neural population responsive to the trigger behavior was related to the planning of following activity.As the neurons which fired before or immediately at behavior onset hardly influenced by the feedbacks from reward or empty reward,they participated in the reward expectation and the drinking behavior accordingly.Specially,neurons with delayed activation were highly exclusive to one of four choice-outcome combinations.These neurons made up 28%±3%and 20%± 3%of total neurons in early and late IGT phase.We performed principal components analysis to activities of neurons with delayed activation,and trained a linear decoder to efficiently distinguish the choice-outcome combination in each trial,which confirmed that neurons with delayed activation were very likely to be the carrier of feedback information on reward type and changes of choice value.In the late IGT phase,the amplitude of activities of neurons with delayed activation in empty reward trials was elevated from 13.18 ± 1.72 to 33.47 ± 5.93(p<0.0001).And the similarity of activities of neurons with delayed activation in empty reward trials also improved from 0.07 ± 0.02 to 0.24±0.02(p<0.0001).These results implied that the importance of empty reward experience was similar to actual reward when calculating choice value.Moreover,the members of each neural population were dynamically shifting while maintaining their function during the whole IGT process.Overall,we established a functional decision-making experiment protocol for mice.Through recording and anslysis of mPFC neural populations’ calcium activities,we confirmed and fitted the relationship between mice’s decision-making behaviors and firing patterns of mPFC neurons.We also revealed the importance of empty reward in the evaluation process.Thus,our results have funded further researches for the machenisms in mPFC coding and mediating decision-making processes. |
