There is a long-lasting debate as to whether mental imagery plays a functional role in cognition: one view states that mental imagery indeed plays a key functional role in cognitive processing, whereas other views suggest that propositional information is sufficient to complete a cognitive tasks and mental imagery is just an epiphenomenon. Recently, an alternative suggestion was proposed to reconcile these two hypotheses, namely that a range of strategies could be adopted to fulfil the demands of a given cognitive task. This thesis was motivated by the general question: What specific strategy could be adopted in cognition in terms of individuals differing in their visual imagery vividness. To this end, mental rotation (MR) tasks were used to investigate individual performance in mental imagery, both in behavioural experiments and event-related potentials (ERPs) studies. Participants were classed according to their vividness of visual imagery (VVI) as individuals with lower and higher VVI. In behavioural experiments, lower and higher VVI individuals were assessed with MR tasks with different task demands. Whereas no group difference were observed in processing simple objects, differential performance was observed in processing more complex visual stimuli: lower VVI individuals were more likely to use piecemeal transformation, whereas higher VVI individuals were found could automatically simplify the task by transferring partial representation of the visual stimuli in their minds' eyes. In the following ERP studies using a standard letters rotation task, longer MR execution time was observed in lower than higher VVI individuals. Rotation-related negativity (RRN) amplitudes were more pronounced in lower as relative to higher VVI individuals. In addition, the onset of RRN was found delayed in lower as compared to higher VVI individuals, while the onset of did not differentiate the two groups. These findings suggest that lower VVI individuals complete the MR process proper later than those with higher VVI. Higher VVI individuals are more able to generate simplified visual representations that can be easily rotated, whereas lower VVI individuals, who are more likely to create less accurate mental representations, might extract and represent additional information from the visual stimuli and therefore produce a longer MR execution time. M.X., an individual who reported the sudden loss of his visual imagery ability was also recruited for an ERP experiment with a standard letter rotation tasks. He performed as well as his age, sex, IQ-matched controls. In addition, he showed the typical linear pattern in RTs as did in his controls did. But differential performance was detected in ERP data. In mirror letters, RRN was detected in the control but not in M.X. This finding suggests that the imagery task could be completed in the absence of the visual representation. Interestingly, in processing normal letters, RRN was observed in both M.X. and the control group, supporting the idea that non-depictive representation could be generated and adopted in MR processing. In turn, this result suggest that RRN is a general index for MR processing, regardless of the format of the mental representation, depictive or descriptive. We then tested our hypothesis from another point of view by investigating whether one could adopt different strategies in MR tasks with different visual stimuli. Specifically, we explored which properties of the visual stimuli are more likely to predict the strategy selection in MR tasks. A behavioural MR experiment with polygon stimuli was conducted aiming at testing the two commonly used strategies used in MR tasks with different types of stimuli. Firstly, the segment number and the number of vertices in polygon stimuli were manipulated to test which property of the visual stimuli is more likely to influence the strategy selection in MR tasks. Secondly, the role of distractors was examined by comparing the stimulus complexity effect in both with- and without-distractors conditions. The results revealed that the number of segments affected the slope of the linear function relating response times to rotation angle but the number of vertices in the polygons did not. This suggests that a holistic strategy was more likely to be adopted in processing integrated objects, whereas a piecemeal transformation was at play in processing multi-part objects. In addition, the stimulus complexity effect was observed in the with-foil condition but not in the without-foil one, providing a direct evidence to support the role of distractors in MR tasks. We then supplemented the investigation of the use of different strategies to comply with MR tasks by further examining if the largely document ageing effect in MR tasks could be explained by strategy selection. Younger and older participants were assessed with a series MR tasks with different task demands. Their visual imagery vividness was controlled. In processing simple objects, the performance of the two age groups was comparable. However, systematic differences were observed between the mental rotation rates of younger and older adults while processing complex objects. Younger participants were faster in processing complex than simple objects, whereas older participants were slower in rotating complex than simple objects. These results revealed that different mental rotation strategies were selected by the two age groups. A simplified representation of the objects was generated and transformed by younger participants in their mind's eyes, while a piecemeal transformation strategy was adopted by older participants. In the ensuing ERP study with simple MR tasks using characters, age-associated slowing was observed in both behavioural and ERPs measures. While the RRN was already present in the early time windows in younger adults (350-500ms), it only emerged in the late time window (500-650ms) in older participants. Consistent with this observation, the onset of the RRN was delayed in older as compared to younger participants. These results suggest that MR processes occur later in older adults and demonstrate that the initial phase before MR proper is one source of the age-related slowing observed in MR tasks. Possible accounts for this age-associated slowing are a prolonged phase of stimulus encoding and/or selective difficulties in directing attention away from the external stimulus towards its internal mental representation. One more study was conducted to explore the general cognitive and neural mechanism underlying MR tasks. To better understand the normal-mirror difference in MR processing, we replicated Núñez-Peña and Aznar-Casanova's ERPs study (2009). The timing course of the planar rotation in normal letters, the planar and non-planar rotation in mirror letters were explored separately for each rotation angle. Moreover, we investigated the temporal relationship between these three processes for each angle. The result uncovered a complex cognitive process underlying MR process with mirror letters in which the non-planar rotation was found to occur at different time as relative to planar rotation for each rotation angle.