By using clips of a TV news show as stimuli in an fMRI experiment, we were able to demonstrate brain activation in healthy participants across a wide age-range that involved the medial and large portions of the lateral temporal lobe - brain regions which have previously been ascribed a crucial role in episodic memory and semantic processing, respectively.
Further, greater activation in left temporal and right frontal cortex was related to older age. Thus, the paradigm applied in this study was not only apt to elicit activation within the declarative memory system, but also revealed a more subtle effect of age-related changes in brain activation during this natural stimulation task. This suggests that research of healthy aging and its alteration in degenerative diseases like dementia of Alzheimer's type might benefit from natural, real-life experimental paradigms.
Natural stimulation paradigms, especially when assessing healthy elderly and demented patients, might be better tolerated by participants as compared to more artificial task settings with rather complex task instructions, and hence, in principle, increase validity of results.
We regard our activation condition as a prototypical example of 'natural stimulation'. This does not refer to similarities between watching TV and real-life interaction with another person. Rather, it refers to watching TV in our culture being one of the core situations designated for information acquisition. As such, TV watching itself has become 'natural' stimulation.
The typical drawback of experiments using natural stimulation as compared to artificial laboratory stimuli lies in the less controlled nature of the cognitive processes evoked. This is especially true for complex, multimodal stimulation. We tried to rule out confounds by contrasting with a control condition which very closely matched to the experimental condition while removing meaningfulness. In fact, control condition stimuli were derived from the exact counterparts of the experimental condition, with the visual presentation rearranged and audio reversed. We thus contrasted brain activity (regional BOLD response) during an episodic memory task, which required semantic processing, to a control condition without semantic or episodic memory processing. This led to almost identical activation patterns when contrasting each of the two with the fixation baseline -- predominant activation of primary visual and auditory cortex. Intensity of visual features was, however, not absolutely matched in the control condition. Coherent motion, e.g., was rather attenuated in the control condition, faces were not recognizable, and more sharp edges were introduced due to frame scrambling. Therefore, differences between experimental and control conditions in early visual areas might be enhanced and not explained by declarative memory. Similarly, while reversing speech (rendering it incomprehensible) reduces semantic processing, at the same time phonological and syntax processing might as well be reduced.
Main Task Effect
The main effect of the memory task (clip > rearranged clip) presumably reflects language processing in lateral temporal cortex, memory encoding in the hippocampus, processing of visual features occipito-temporal cortex, and multimodal integration in the STS.
We assume the observed left-lateralized temporo-lateral activation to correspond largely to the ventral stream brain regions which have been demonstrated to be crucially involved in language processing [14, 15]. Dorsal stream regions seem to be less engaged in the task, presumably due to high demands on input (comprehension) in contrast to output processing (articulation). The areas involved include the left temporal pole, which has been termed the 'semantic hub'  as it is implicated in semantic processing [17, 18]. The inferior, posterior temporal activated region presumably corresponds to a sound-meaning interface, a function which has been ascribed to this region  that has also been termed the 'basal temporal language area' . Further activated spots along the long axis of the MTG map nicely onto what has recently been termed the large-scale semantic network of the human brain . We thus assume that much of the observed left lateral temporal cortical activation pertains to semantic processing of incoming information, which is a prerequisite for episodic memory formation provided by the hippocampus . The crucial role of the hippocampus for episodic memory formation and, in part, retrieval is well characterized in the literature . This suggests that the observed hippocampal activation reflects episodic memory-related processing.
The link between memory task-related activation patterns and declarative memory processing is corroborated by the finding that stronger task-related activation was related to better performance in temporal and frontal cortex (see Results).
It has previously been demonstrated that complex, multimodal stimuli, like watching a sequence of a movie, are apt to detect activity of several cortical regions which are specialized for unimodal features like motion, color, or faces . The task-related activation pattern in the current study as well includes cortical regions which are specialized for processing unimodal features like motion (in the occipito-temporal junction, presumably V5), or faces (in the posterior fusiform gyrus; c.f. Figure 1). Beyond signal responses to unimodal stimulus features, the main task effect in the current study as well evoked activation of multimodal integration cortex, the STS .
Brain regions which exhibited an increased task-related activity with older age were observed in left temporal, frontal, and posterior parietal cortex, as well as in a right frontal region, corresponding to BA44. This increased BOLD response in the latter region might reflect a compensatory mechanism in terms of recruitment of an additional contralateral homologue of the left hemisphere Broca area for language comprehension, as has been observed in left hemisphere stroke patients . Likewise, increased frontal activity with age has recently been described , though coupled to a decrease in occipital cortex, which we did not see in our data. It might as well point to a reduced hemispheric asymmetry in older age, which has previously been hypothesized . Alternatively, frontal cortex involvement in elderly participants might as well reflect a reduced inhibition of an articulatory network , which, although tapped by speech input, is rather irrelevant for task completion.
The observed greater activity concomitant with older age in left superior temporal regions -- neighboring those areas recruited by the task -- might be driven by increased effort, which elderly participants needed to perform as good as younger participants (the regression analysis controlled for performance effects). Left mid--part superior temporal cortex has recently been associated with semantic processing .
Furthermore, a left posterior parietal region -- close to the intraparietal sulcus -- displayed greater activity in the elderly. This might reflect increased involvement of the dorsal attention network  or, more specifically, attention to memory [29, 30], which was not required for successful task completion in younger participants.
A theoretically possible cause of the observed age-effect on activation might be altered hemodynamics in elderly participants, unrelated to cognition. In healthy elderly, a decline of cerebral blood flow has been shown recently , as well as reduced task-related fMRI BOLD signal . In contrast, in our study an increased BOLD response with increasing age was observed, which suggests a neural rather than vascular cause of the observed effect.
It was attempted to rule out contribution of impaired hearing or vision in elderly participants to age-related activation effects. Each participant confirmed good hearing and vision of stimuli before the experiment started. However, as this was not formally tested, it should be considered a possible confound in this analysis.
Activation decrease associated with increasing local gray matter atrophy, has been reported previously . The fact that this was not observed in our study suggests that healthy elderly participants did not have sufficient atrophy to reduce the BOLD signal in any region.
In contrast to the robust main task effect, the observed age-effect is considered a secondary, preliminary result of the current study that requires replication by studies assessing larger numbers of participants.