Compensatory cognition in neurological diseases and aging: A review of animal and human studies

Abstract

Specialized individual circuits in the brain are recruited for specific functions. Interestingly, multiple neural circuitries continuously compete with each other to acquire the specialized function. However, the dominant among them compete and become the central neural network for that particular function. For example, the hippocampal principal neural circuitries are the dominant networks among many which are involved in learning processes. But, in the event of damage to the principal circuitry, many times, less dominant networks compensate for the primary network. This review highlights the psychopathologies of functional loss and the aspects of functional recuperation in the absence of the hippocampus.

Introduction

One of the outstanding characteristics of the human brain is its ability to learn and adjust behavior based on experiences and memories. To produce such complex behaviors, the brain continuously undergoes structural as well as functional re-organization [177]. Brain plasticity is the ability of the neural network to rearrange its connections and functions, both at the system (circuit) and molecular (synaptic) levels in response to internal or external inputs [142]. For instance, learning-induced plasticity involves activation of postsynaptic NMDA glutamate receptor (NMDA-R) induced influx of Ca²⁺_, followed by downstream signaling via secondary messengers, ultimately triggering long-term structural and functional changes in the neurons [109], [147], [230]. Extensive training or learning re-organizes the neural circuitries/network and helps the brain to store and retrieve the learned information for a longer period. If the affected circuitry is damaged, it results in a lifelong cognitive deficit. But interestingly, some of the lost abilities can be regained and compensated [206]. In this review, we are highlighting the events of structural and functional recovery and compensatory cognitive ability after the loss of the principal network.

Memories are encoded via a dedicated group of neuronal populations, which are initially active at the time of acquisition, and re-activated in their later phase at the time of retrieval [170]. According to the multiple memory system’s theory, distinct circuits are specialized to perform specific functions [160], [68]. These circuits do, however, engage in competition, with the dominant one performing the specific function. Studies suggest that synapses undergo synaptic scaling, which provides overall weight normalization at the neuronal level. As a result, competition is generated between different neuronal populations [234], [241], [230]. Different situations, such as perceptual categorization, action selection, decision-making, and attention, have all been thought to involve competitive selection [168]. Fear learning is the most well-known instance, which is mediated through multiple pathways (primary and alternative pathways) [68]. Fear-conditioned learning is dominated by the primary ‘dorsal hippocampus’ pathway. It normally sends inhibitory signals to the alternative prefrontal cortical pathways. These alternative pathways, however, are free from inhibitions in the absence of the primary pathway and hence mimic its compensatory functions, albeit less effectively.

The process of memory encompasses sequential events for acquisition, storage, and retrieval of new information. The physical and neural imprints of memories are encoded in the form of engrams [179], [212], [63], [115], [149], [95]. Conventionally, memory is broadly classified into long-term memory and short-term memory. Long-term memory is further divided into (A) declarative memory (i.e., explicit knowledge of facts and events), which is further sub-divided into (a) episodic memory (spatial and context-related memories) and (b) semantic memory (memories of facts and figures), and (B) Non-declarative memory (implicitly acquired, of unconscious knowledge) involves procedural learning, priming, etc. [20]. A broad classification of different memory systems is shown in Fig. 1. Short-term memory comprises working memory, which includes temporary storage and manipulation of sensory information for a very brief period [189], [10], [148]. The classical conditioning and non-associative forms of memory, including skill-based knowledge, develop gradually but with little ability to report what is being learned [68]. The consolidation of declarative memories involves the hippocampus and structures of the medial temporal lobe (MTL) for the processing of memory information during its acquisition and consolidation. On the contrary, non-declarative memories rely on structures like the neocortex, amygdala, and cerebellum. Further, within the neocortex, different sub-regions are thought to underlie different types of semantic knowledge [157], [156], [227]. For instance, knowledge of words and faces is mediated by the areas in the fusiform gyrus, whereas knowledge of the higher concepts is handled by anterior temporal regions [157], [27], [12]. The hippocampus is believed to be the central coordinator of memory processing as it combines an event’s spatial, temporal, and contextual attributes to form a conjunctive representation [62], [63], [41]. To perform this, it acquires the information either from a single network or multiple brain areas to stabilize a specific memory as a neuronal engram [67], [179], [212], [116]. The engram cells play an essential role in the stable representation of memory [231]. For instance, contextual fear conditioning is a learning paradigm used in animal models to study associative memory. When a neutral stimulus (context or tone) is paired with an aversive stimulus (e.g., electrical shock), the animal shows a conditioned response, i.e., freezing even in the absence of electrical shock. The attributes of the contextual and aversive stimuli are acquired individually by separate networks in different brain areas, the association is, however, formed in the hippocampal neurons [155], [154], [61]. The hippocampus acquires contextual representation, and the associative connections between the hippocampus and amygdala are strengthened in response to the electrical shock. Such processes suggest that the encoding simultaneously takes place in multiple memory-associated brain areas [171], [67], [169], [198], [13], [86]. Damage to these areas causes profound memory impairment leading to both retrograde as well as anterograde amnesia.

Fig. 1.

A broad classification of memory. Based on the attributes of memory retrieval patterns (short-lasting and long-tasting memories), memory is classified into two categories. Similarly, long term-memory is classified into two broad categories based on the active and passive nature of recollection.

There is very little possibility that cognitive function will improve following neuronal injury. In prodromal and early-stage neurodegenerative and neuropsychiatric illnesses, the patients can function at a level that is comparable to the general population while showing clear evidence of cortical and subcortical loss. This upkeep of typical performance is thought to result from the initiation of compensatory mechanisms, or adjustments in brain activation within a network specialized to a task, or in the recruitment of a region outside of the task network. The subject may improve their cognitive functions either (a) through behavioral adaptation and/or (b) by the development of compensatory neural pathways in the brain.

Behavioral compensation

Whether the behavioral alterations are merely the result of compensatory methods used in response to a damaged neurological system remains a crucial question. However, studies indicate that there might be a connection between subjective health and compensatory skills. [77], [129]. The compensatory strategies help the subjects to learn specific methods and skills to compensate for the cognitive deficiency and certain case reports highlight such compensatory strategies to improve cognitive functions. [78].

The patient JC: a curious case of behavioral compensation

The concept of memory compensation after the damage of the memory-associated areas dates back to a curious case of patient JC [247]. JC suffered from left cerebral artery aneurysm due to subarachnoid hemorrhage in an accident at the age of 20 years. After the accident, he was diagnosed with selective memory impairments (anterograde amnesia), along with severe epileptic seizures. During his clinical assessments, based on general IQ such as naming, reading, and visuospatial; perceptual & executive functions, he was able to score remarkably well. But, his scores were not comparable to those of normal healthy adults when assessed on memory tests/scales. Interestingly, he was able to remember events that preceded the accident during the retrograde amnesia tests. Also, his immediate memory seemed to be intact. JC’s initial treatment approaches to improve the condition included using mnemonics diaries, rehearsal strategies, and even splitting up a long task into small steps. These treatment strategies helped him to retain some information, but he still faced problems with routine memory tasks.

Gradually, he started following the habit of making notes and continued it for several years. For instance, he started making lists of his accounts and keeping dictaphone recordings for things he needed to remember in the near future. For example, telephone calls, things he had to buy, etc. This way, he was finally able to develop a compensatory system that could enable him to survive independently. Since his implicit memory was seemingly intact, he used it to develop skills that could help him compensate for other memory impairments.

JC developed ‘the system of compensatory memory’ into a foolproof system. He modified his system of compensatory memory many times. Even after so many modifications, in a common men’s view, it appeared to be a herculean task to make JC’s system of compensatory memory a foolproof system. However, it cannot be denied that JC had developed a very effective system to recall the partially consolidated/labile memory of the amnesic subjects. It is intriguing that JC was able to recall some of the memories, but under similar conditions, others could not. Later, Wilson and Watson [248] proposed that younger people (∼30 years of age) can develop compensatory memory relatively in better way than the aged subjects [248]. They also observed that the amnesic subjects without additional cognitive deficits developed memory compensation more adequately than the subjects having additional cognitive issues [248].

Neural compensation

Studies have demonstrated that in the events of brain damage, a neural compensation process also occurs. For example, the bilateral labyrinthectomy severely impairs postural stability but gradually with time the animal can stand unsupported, ataxia profoundly reduces, and locomotion speeds gradually increase [159]. Further, neuroimaging data show that the restructuring of cerebral resources in older persons may help them make up for cognitive and brain function deficits [114]. Similarly, in MRI findings on neurodegenerative disease, compensation is typically shown in the early stages of a variety of conditions, such as Alzheimer's and Parkinson's disease [88]. Compensatory mechanisms develop in older subjects and subjects suffering from neurodegenerative diseases, however, we have to have a clear understanding of how it can develop.

Development of compensatory pathways in neurodegenerative disorders

Alzheimer’s disease (AD)

Alzheimer's disease (AD) is the leading cause of dementia and cognitive decline in older populations. It is mainly characterized by the formation of amyloid β (Aβ) plaques and neurofibrillary tangles, exhibiting substantial loss of synapse and spine density [219]. Selective damage in brain areas involves the neocortex, hippocampus, amygdala, and limited regions of the medial nucleus of the thalamus, locus coeruleus, frontal & cingulate cortex, etc. [246], [214], [223]. Several studies have reported gradual and progressive deterioration of episodic memories [167], [210]. In the animal model of AD, various studies have demonstrated the compensation in terms of either cognitive performance or associated pathology. For example, Billings et al. have shown that training of spatial learning tasks such as longitudinal water maze produces a significant but transient improvement in learning other tasks, as well as a reduction in the deposition of amyloid-beta and tau proteins [19]. In the triple-transgenic mouse model of AD (3 × Tg-AD), the mice were trained and tested at 3-month intervals from 2 to 18 months. The performance improvement was observed after six months. Interestingly, it further delays the redistribution of amyloid-beta proteins to extracellular plaques and reduces the deposition of its oligomers associated with cognitive decline. Such a study demonstrates that repeated spatial training can significantly delay the development of neuropathology and a decline in spatial memory. Similarly, in another study, Jankowsky et al. reported the role of environmental enrichment in the alleviation of Alzheimer’s-associated learning and memory deficits. Intriguingly, AD mice showed a significant improvement in cognitive performances with environmental enrichment compared to the control group of animals [111]. Further, in another study, transgenic mice overexpressing Swedish doubly mutant amyloid precursor protein(APP), housed in the enriched environment were found to have significant global improvements in cognitive functions when compared with those put in standard housing for four months [7].

Mild cognitive impairment (MCI)

Studies on human subjects with mild cognitive impairment (MCI) have shown hyper-activation in the hippocampus, posterior medial temporal, and connected fusiform regions during visual memory tasks when compared to the older control group [57], [94], [125]. Another study using a similar methodology instead showed higher activation of the dorsolateral prefrontal cortex (dlPFC) and posterior parietal cortex (PPC) in patients with MCI when compared with cognitively normal age-matched controls. In a SPECT study [30], voxel-based morphometric analysis was performed to obtain brain images of patients with incipient AD to investigate the relationship between grey matter (GM) atrophy and functional alteration (in terms of hypo-perfusion). On masking the atrophic area maps with those of the non-significantly hypo-perfused ones, they observed the atrophic areas (grey matter, white matter, or CSF) without significant functional deficits, which they referred to as functional compensatory areas, and the areas were posterior cingulate cortex, amygdala, the dorsal hippocampi, and the insular cortex.

Huntington’s disease (HD)

Huntington’s disease (HD) is an autosomal dominant neurodegenerative disease marked by chorea (involuntary, irregular unpredictable body movements), cognitive issues, and mood abnormalities, especially depression [203]. Initially, neurodegeneration begins in the dorsal striatum and later affects ventral striatal areas to some extent. Malejko et al. investigated the possibility of functional compensation of early neurodegeneration in prodromal Huntington's disease using structural and functional MRI [150]. They reported that patients with HD exhibited structural changes in the dorsal striatal area with marginal changes in the volume of the ventral striatal area. The patients showed prolonged reaction time in the motor response tasks, however, behavioral performance in the rewarding task was intact. Interestingly, the reward-associated fMRI signaling in the HD group was significantly different among the bilateral sides in the ventral striatum, orbitofrontal cortex/anterior insula, and other reward-associated areas [150]. Another PET study has demonstrated that the activation responses during learning significantly increased in the left mediodorsal thalamus and orbitofrontal cortex in pre-symptomatic HD patients [72]. However, Klerman et al observed that brain atrophy enhanced performance-related activity in the right parietal cortex along with the increased functional coupling between the right prefrontal cortex and a left hemisphere network [127]. This lateralized activation could be attributed to its pathophysiology where the left hemisphere is shown to be more susceptible to neurodegeneration [134], [251], [127], [165]. All these demonstrate the development of active compensatory processes in premanifest-HD, which could be attributed to their increased cognitive demands [196], [83], [87], [150], [127], [252].

Development of compensatory pathways in neuropsychiatric disorders

Obsessive-compulsive disorder (OCD)

Evidence suggests that patients with OCD may experience difficulties in the formation of working memory [54], [33] and some other cognitive functions [35], [201]. Working memory primarily depends on the frontoparietal network. However, de Vries et al. have reported hyperactivity in left dlPFC, left premotor/pre-supplementary motor area (SMA), and left precuneus areas after visuospatial working memory task when compared to healthy participants [52]. Another study also compared patterns of brain activity using fMRI in patients with OCD and healthy controls during spatial and verbal memory tasks [102]. Hyperactivity was observed in the regions of the left inferior frontal cortex, the middle part of the left inferior frontal sulcus, the left inferior frontal junctional (IFJ) area, and the intraparietal cortex. These brain areas are responsible for the articulatory rehearsal and phonetic maintenance of verbal information [90], [91], and the same areas have been reasoned as the compensatory areas to overcome working memory deficiency in patients with OCD.

Posttraumatic stress disorder

Post-traumatic stress disorder (PTSD) is characterized by recurrent memories from traumatic events in the past. It is a robust negative memory and, once acquired, remains for a lifetime. The life of an individual becomes miserable with time. Brain structural aberrations, such as a reduction in the hippocampal and ventromedial prefrontal cortex (vmPFC) volumes, have been found in PTSD patients [24], [92], [228]. A few studies have reported that PTSD is associated with selective volume loss of the CA3 and dentate gyrus subfields in the hippocampus [244]. It may be causing the inability to acquire context information, which is thought to be mediated through the dorsal hippocampus [120]. Nevertheless, studies have also reported the role of the basolateral amygdala (BLA) in the acquisition and expression of fear memory, which works in concert with the DH and vmPFC [11]. The dorsal hippocampal lesion causes a profound memory deficit [120], while the BLA seems to be playing an essential role in consolidating the fear memories in other brain areas [69], [153], [50], [186]. It is, therefore, likely that the hippocampus may encode the contextual information by forming an association with the aversive information, shock, for instance, by interacting with the BLA.

In rodents, lesions of the dorsal hippocampus ubiquitously extinguish context fear memories. However, animals whose hippocampus has been damaged before conditioning display the unique ability to overcome hippocampal loss, the mechanism of which remains unknown. Evidence for this neuronal compensation after the dorsal hippocampal loss was primarily reported by Wiltgen et al., wherein the formation of compensatory circuits after DH damage through training up to 16th-day post-lesioning was reported [249]. After the complete excitotoxic DH lesion, five training trials induced normal levels of contextual learning in rats. The study suggested the role of the neocortex in the acquisition of contextual fear memories in the absence of the dorsal hippocampus. However, the rate of learning is comparatively slower than the hippocampus [249]. The neocortex is thought to be the putative hippocampal independent learning area, albeit less efficient, since the neocortex, in concert with the hippocampus, drives the process of learning in intact mammals [179]. Similar results have also been observed in another study [137]. Although there is sufficient evidence for the compensation taking place for DH damage, the exact site and the nature of the compensation have not been explored yet. To determine its temporal aspects and whether the compensatory learning is as stable as that of the intact hippocampus, excitotoxic lesions of DH were performed in rats using a similar paradigm in another study. Rats were tested for contextual fear conditioning on the 1st, 3rd, 10th^, and 30th-day post-training, to observe the temporal changes in fear expression. The results showed that the expression of contextual fear memories formed after the hippocampal damage diminished over time [254]. It is likely that the compensatory memories could be volatile in nature or a short-lived event. Interestingly, Fanselow has proposed an alternate hypothesis that the compensatory circuits form contextual memories only in the absence of the hippocampus but not in the presence of the hippocampus [68].

A recent study has hypothesized the recruitment of the infra-limbic (IL) and pre-limbic (PL) prefrontal cortices for compensatory memory following hippocampal damage. The PFC has a well-established role in the storage of long-term contextual fear memories. In this study, post-training lesions of the DH in rats showed a significant freezing response, which was no different from the sham group. In addition, the DH + IL or DH + PL lesioned rats were also examined to see the individual roles of these sub-regions, which were found to have completely lost the compensatory context fear in both conditions. This concluded that both IL and PL are individually necessary for the formation of compensatory circuitry after DH damage. In addition, the compensatory context memories also correspond to an increase and decrease in IL and PL activities, respectively [255]. The study was further extended to see whether a cross-talk between IL and PL results in the development of compensation for the DH. The DH + IL lesioned animals did not show any compensatory context-specific fear. Contralateral lesions of IL and PL and subsequent testing for contextual fear showed the progression of compensatory contextual fear conditioning [255]. The IL and PL have contrastingly different roles in fear expression [84], [239], and also send their connections to different sub-regions within the amygdala [238], [79], The medial prefrontal cortex may be a prospective site for the compensation of DH primarily because of its role in the long-term storage of remote contextual fear memories. Hence, it is likely that when the DH is intact, the prefrontal cortex and amygdala may be receiving the contextual fear memory along with the DH. Since the DH network could be acting as the primary and dominant network for the formation of an association of fear and context memories, the subsidiary network such as mPFC may not be that strong initially for the formation of the association in the presence of the DH. Nevertheless, in the absence of the DH, the mPFC network must have gained the property to compensate for the loss of DH. Very recently, it has been reported that memory deficits in hipocampus lesioned mice could be rescued using the cortical application of induced synchronized oscillatory signals (iSOS) in a contextual fear memory task. The freezing levels of control mice were found to be similar to those of the hippocampal-lesioned rescued mice [146]. The neural circuitry for contextual fear conditioning is thought to involve the dorsal hippocampus as the primary area. Individual contextual (neutral and aversive) inputs are separately acquired in different somatosensory/association cortices, from where these inputs are relayed onto the DH, as well as to the amygdala through the entorhinal cortex (EC) [211], [37], [34], [124]. The DH integrates these inputs to form a consolidated memory, which is later on transferred to higher cortical areas for long-term storage. In the absence of DH, the inputs coming from EC will no longer be able to relay the information to the DH. We reasoned that in such conditions, an existing connection between EC and mPFC will be strengthened and the CS-US association will be formed in the mPFC area in the absence of DH (Fig. 2).

Fig. 2.

Possible contextual and compensatory contextual fear conditioning circuitries in the brain. The sensory signals ‘conditioned stimulus’ (CS) and ‘unconditioned stimulus (US) are acquired individually in somatosensory cortices, dorsal hippocampus (DH), and basolateral amygdala (BLA). The BLA through the periaqueductal grey area helps in the induction of freezing behavior during memory retrieval. The somatosensory cortex project to the DH via the entorhinal cortex (EC). On the other hand, EC also projects to the medial prefrontal cortex (mPFC) which includes infralimbic (IL) and prelimbic (PL) areas. In the absence of DH, the CS and US signals will no more be able to integrate into the DH. Rather the EC may be able to send the CS & US signals to the IL & PL with multiple training trials. Hence, in the absence of DH, the development of EC-mPFC circuitry with time may help mPFC to take over the role of DH as a compensatory contextual fear conditioning center. Arrowheads denote neuronal projections from one area to another. The thickness of the arrows depicts the strength of connections between those areas. Black arrows show active projections. Grey arrows show damaged/inactive projections.

In parallel reports, the BLA lesions during the pre-training period abolish fear acquisition [191], [101], [80]. But, this loss of fear conditioning can be regained via excessive training [193], [196], [222]. This again supports the view of the existence of compensatory pathways, which take over the functions, albeit less efficiently. Another study has reported that the Bed nuclei of stria terminalis (BST) compensates for the excitotoxically damaged BLA. The freezing levels after training in BLA-damaged rats were similar to those of the intact rats [199]. Also, when BLA was lesioned along with BST, the freezing behavior was significantly reduced, suggesting that context fear acquisition was lost in the absence of BST [196].

Development of compensatory pathways in the aged brain

Apart from the causes due to underlying disease conditions, cognitive deficits have also been prevalent in the geriatric population. However, it is still uncertain whether cognitive functions necessarily decline with increasing age. The brain size, vasculature, and cognitive ability vary as we become older. With advancing age, the brain shrinks, and changes occur at every level from molecular (biochemical synthesis and release) to morphological as well as physiological levels.

Volumetric changes in the brain with aging

Aging induces structural, physiological, and behavioral changes in the brain. Aging commonly comes with cognitive decline [26], though it can’t be generalized [74]. The magnitude of cognitive decline varies for different cognitive processes. For example, the executive functions are typically the first to show impairment during normal aging, language abilities remain relatively intact [163], or even improve in terms of vocabulary, semantics, and speech processing [122], [237], [9].

Structural alterations have been universally reported in the form of gray and white matter atrophy [4], [40], [110], [97]. An fMRI study used voxel-based morphometric analysis to investigate differences in grey and white matter tissue volumes [70]. A significant reduction was found in the case of grey matter volume in the older population as compared to the younger ones, predominantly in the frontal, cingular and insular cortices. However, limbic and para-limbic regions were spared. Additionally, when compared between males and females separately, there was a steeper decline in grey matter volumes in females when compared with the males in both younger and older populations. On analyzing the white matter volumes, again, a negative correlation was found, mostly in the thalamic regions, in the older population as compared to the young ones. Apart from cortical areas, grey matter atrophy has also been reported in cerebellar regions [1], [2], [202], some of which are involved in cognition (language-specific and executive functions) rather than motor functions [25]. However, no significant gender differences were observed in this case. Similar patterns of decreased grey and white matter volumes with advancing age have been observed in other studies [89], [47], [119]. Although aging is known to be associated with grey matter atrophy, the degree of this atrophy is heterogeneous across the elderly population [73].

Alterations in neurotransmitter signaling with aging

Aberrant neuronal network activity in terms of altered neurotransmission is commonly reported in the aging brain. The excitatory neurotransmitter ‘Glutamate’ plays an essential role in memory consolidation through the process of long-term potentiation. Age-related decline in glutamate neurotransmission has been well reported [107], [130]. Aging-associated decreased glutamate levels have been observed in the hippocampus and areas of the cerebral cortex in rodents [218], [256], [93] and also in humans [216], [118], [253]. Moreover, it has been shown that this decline in glutamate levels is steeper in males than in females of the same age group [36]. Another study using proton MR spectroscopy investigated age-related changes in glutamate and glutamine levels in the brain between normal and aged populations. Reduced glutamate levels were observed in the grey matter motor cortex in the older population when compared with the younger subjects. At the same time, higher concentrations of glutamine were observed in the white matter corona radiata in the older population [118]. A recent study found decreased glutamate concentrations and significantly higher concentrations of glutamine in the older population compared with younger adults [209]. The higher concentrations of glutamine could imply impaired glutamate metabolism due to disequilibrium in the astroglial neurons since glial cells are the sources of glutamine. In addition, a decrease in the number of glutamatergic neurons has also been observed in aged rats in the cortical regions [104].

Another neurotransmitter, acetylcholine, is major neurotransmitter of the cholinergic system involved in the regulation of cortical activity [82]. Deterioration of cholinergic neurons with aging progressively leads to learning and memory deficits. Further, the enzymatic activity of Choline acetyltransferase (ChAT) (required for the synthesis of acetylcholine) in the hippocampus and cortex reduces progressively with age [189], [53]. In contrast, there are also reports which have found no significant difference in ChAT activity in the cortex as well as caudate-putamen region between younger and older populations [85], [44]. In addition, the muscarinic, as well as nicotinic receptor binding, has also been reported to be reduced in the hippocampus, entorhinal cortex, caudate, putamen, as well as other cortical areas [208], [45], [100].

Dopamine is essentially involved in the regulation of movement and control functions. It is also a major contributor to reward learning, motivation, and behavior. Age-related physiological alterations have been shown to reduce the number of dopaminergic neurons in the substantia nigra region, both in rodents [81], [174] and in humans [205], which is associated with motor impairments. In contrast, few reports demonstrate no significant differences in the number of dopaminergic neurons between control and aged subjects [71]. Deficits in dopaminergic signaling have also been associated with a decrease in several neurotrophic factors like brain-derived neurotrophic factor (BDNF), nerve growth factor (NGF), and other neurotrophins [221]. Loss of these trophic factors has been shown to augment the onset of neurodegenerative disorders [184].

Several studies using MR spectroscopy revealed decreased GABAergic signaling in aging brains in both animal models [96]) and humans [143]; particularly the one mediated by GABA-A receptors [98], [207], [161], [195]. In addition, other neurotransmitter systems such as serotonergic, noradrenergic, etc. are also affected in the aged brain in terms of hyper-excitability and cytotoxicity [28]. An aging brain does not significantly lead to the deterioration of neurotransmitter signaling. Hence, not all functions are completely lost. However, a dysfunctional system of neurotransmission may augment the process of neurodegeneration and increase the vulnerability to age-related neurodegeneration [82].

Structural and morphometric neural alterations with aging

Several studies report a decrease in arborization of the dendritic trees of pyramidal neurons, mainly found in the prefrontal, superior temporal, and pre-central gyrus in humans [172], [51], primates [183], [58], [117] and mice [220], followed by loss of dendritic spines [185] and reduction in the number of oligodendrocytes [188], [65]. Disorganization of the myelin sheath along with axonal reduction has also been observed, which directly affects the conduction velocity. The altered action potential of neurons due to improper axonal insulation ultimately leads to cognitive and behavioral decline [190], [112], [22]. In addition, the role of glial cells in aging (other than their committed roles in maintaining synaptic homeostasis, glycogen storage, and synapse development) have also been emerging in reports [225].

Compensation for age-related neurodegeneration

Evidence of an age-related decrease in occipital activity with a substantial increase in frontal activity [also known as “posterior-anterior shift in aging” (PASA)], has provided models of compensatory cognition through various neuroimaging techniques. Few studies [140] investigated memory-dependent decision-making in younger and older populations using fMRI studies. Additional activation of the vmPFC apart from the regions of the frontoparietal network (dlPFC) was significantly higher in older than younger adults in a delayed vs intermediate memory choice task. Another fMRI study analyzed the encoding of face-name associations of healthy individuals. They found that the right hippocampus showed stronger neuronal activity during tasks of low as well as high encoding demands compared to baseline conditions in tasks. In contrast, the left hippocampus showed higher activities only in tasks requiring intermediate & high encoding demands. It was hypothesized that the left half might be an additionally recruited area to avoid a sharp decline in memory performance of the hippocampus of the right hemisphere. Some additional MRI studies have also reported an increase in regional cerebral blood flow in inferior frontal and anterior temporal cortices in early AD patients as compared to normal [145], [48], [140]. However, whether this increase correlates with the increase in cognitive functions is not known. While functional compensation is evident even after substantial damage to the memory systems, the reports are only limited to mild or no impairments. Nevertheless, the above reports are in line with the present hypothesis that in the absence of a brain circuit underlying a particular memory function, alternate neuronal circuities may be recruited to perform the functions.

Development of compensatory pathways and sleep

Role of sleep in memory consolidation and compensatory memory

Sleep is beneficial for the brain and body, and enhances memory consolidation and retention. Even a 6 min daytime nap helps facilitate memory [233], [128], [133]. It is known that during sleep, there is a replay of the firing pattern of the place cells as observed in the rats, similar to the one seen during active encoding. Also, studies have reported neuronal reactivation through the firing of similar patterns but with higher frequency during sleep [178]. These reactivations are generally observed in the hippocampus as sharp-wave ripples (having field potential oscillations of about 180 Hz) [56] during slow-wave sleep (SWS). SWS, in particular, is beneficial for the consolidation of declarative memories. In an fMRI study, it was found that hippocampal areas that were active during a spatial working memory task in subjects were reactivated during NREM sleep [187]. However, some studies have reported that this neuronal replay occurs during wakefulness as well. Such processes are often referred to as offline consolidation. [121], [162], [226]. Therefore, it is evident that sleep plays a permissive role in memory consolidation.

Sleep architecture may not necessarily change uniformly after learning different types of tasks. For example, cued fear conditioning is a sleep-dependent task but it may not require robust augmentation in sleep amount as such. Although, sleep architecture changes in fear-conditioned animals, NREM sleep amount increases while REM sleep amount decreases during 24 h of sleep-wake (S-W) recording after fear conditioning [113], [99], [131]. Nevertheless, such changes in sleep architecture did not occur in compensatory contextual memory (in the absence of DH). REM sleep remains suppressed after cued-fear conditioning [113], [131] and CxFC [200]. We have reported that REM sleep decreases after CxFC in DH-non-lesioned animals, even after multiple conditioning [120]. Moreover, we have recently reported that after cued-fear conditioning, REM sleep exclusively decreased in the consolidated memory group but not in the memory-impaired group [132]. We observed that DH-lesioned rats could express context fear after the second conditioning; nevertheless, it did not alter their REM sleep amount. No change in REM sleep amount in the DH-lesioned rats suggests that it might be related to the consolidation of fearful memories in some alternative brain circuits, such as the medial prefrontal cortex (mPFC), which may not influence sleep architecture.

Sleep disturbances and aging

It was observed that most of the age-related changes in sleep occur after the age of sixty [182]. With aging, the architecture of sleep shifts from the conventional characteristics in terms of prominent changes in the duration of sleep, as well as different vigilant sleep states, to micro-level changes like quantity/quality of sleep [152]. Amount of slow wave sleep (SWS) and K-complexes during NREM sleep stage, total sleep time, and sleep efficiency, as well as latency, reduced with aging [166], [138], [75], [158], [59]. Also, increased wakefulness during the nighttime has been prevalently reported [245], [236], [204], [126], [43], [240]. The medial prefrontal atrophy in the older age group has been correlated with disturbed sleep in terms of reduced SWS [151], which also led to the impairment in hippocampal-dependent memory [151]. Naps during the daytime have been considered a normal routine in most cases. However, this phenomenon is more prevalent in the older population as compared to the younger ones [76], [164], [66].

The putative neurobiological mechanisms underlying sleep alterations in aging involve neurodegenerative or neuropsychiatric disorders, sleep disorders, hormonal imbalances, or other abnormalities associated with aging [144]. It has been reported that a decrease in the release of galanin neuropeptide from the ventrolateral preoptic/intermediate nuclei has been correlated with advancing age [141]. In addition, a 40 % reduction in orexin-expressing neurons has been reported in older rats as compared to younger ones [123]. Sleep is predominantly regulated by circadian rhythmicity and homeostatic drive, which is further regulated by extracellular adenosine, a metabolic byproduct that accumulates while we are awake [60], [108]. Age-associated changes in adenosine production, as well as adenosine receptor levels, have been reported [38], [31], [213]. Hence, these as well as many other components could be attributed to the cause of sleep disturbances in the aging population.

Role of oscillatory waves of the brain in compensatory memory

Sleep-associated oscillatory waves such as hippocampal theta oscillations during REM sleep and gamma waves play a crucial role in the consolidation of memories. Polysomnographic studies demonstrate that theta activity remains elevated during consolidation as well as retrieval after context-dependent memory tasks [194], [5], [105], [23], [39], [46]. Overnight alterations in theta coherence between the hippocampus, medial prefrontal cortex, and amygdala during paradoxical sleep were specifically linked with changes in fear memory and thus it has been proposed that coherent theta activity in the limbic system promotes fear memory during paradoxical sleep [194]. Further, optogenetically induced theta rhythms (7 Hz) in the CA1 area rescued the consolidation of contextual fear memory in the sleep-deprived animals [180]. These studies suggest that REM sleep components such as theta waves during REM sleep plays an essential role in the consolidation of fear memories.

In addition to theta activity, gamma activity (25–100 Hz) has also been recognized for its role in cognitive function, including memory. It has been discovered that both human AD patients and rodent AD models exhibit aberrant gamma oscillations. Gamma entrainment has been demonstrated in recent studies to considerably lessen AD pathology and improve cognitive function in AD-prone mouse models through auditory and visual sensory stimulation. According to the initial evidence from AD patients, gamma entrainment therapy can alleviate several degenerative disease markers, reduce brain shrinkage and loss of functional connectivity, and enhance cognitive function [232].

Development of compensatory pathways and adult hippocampal neurogenesis

Increased neurogenesis from multipotent neural stem cells in the subgranular zone of the hippocampus promotes hippocampal-dependent learning [136], [15], [16]. Further, its suppression leads to memory impairment [55], [103], [224]. Additionally, impaired adult neurogenesis has been observed widely in various animal models of neurodegenerative disorders [250], [217], and in aged animals [8]. Few studies have proposed that environmental enrichment and exercise, both act as strong neurogenic stimuli, and may have therapeutic benefits to minimize the adverse effects of aging, and neurodegeneration [106], [243]. However, the role of adult neurogenesis in compensatory cognition of neurodegeneration is not known. Interestingly, in a recent study on rats, it was observed that one side hippocampectomy induced compensatory neurogenic response in the opposite hippocampal region. It was proposed that it may have been the reason for a considerable absence of a cognitive deficit in human patients who underwent unilateral hippocampectomy to treat refractory temporal lobe epilepsy [29]. The lack of cognitive deficiency in the patients of unilateral hippocampectomy could be attributed to the increased neurogenesis in the opposite hippocampus as a compensatory approach. Nevertheless, future studies await to provide in-depth knowledge of the role of adult neurogenesis and the development of compensatory processes.

Influence of brain and cognitive reserve on the development of compensatory pathways

JC’s case explained above was undoubtedly extraordinary, wherein he enabled himself to develop behavioral compensatory systems to live independently. Although, it is not known whether his compensatory behavior induced the neural circuit reorganization to allow neural compensation. However, numerous reports have shown that behavioral adjustments may induce cytoarchitectural changes including an increase in synaptic strength and synaptic turnover. A concept of cognitive reserve has been proposed which explains why there is a difference between a person's measurable level of brain pathology and expected cognitive performances. The brain reserve is a person's unique sensitivity to brain damage, which mainly depends on two factors: (a) the severity of the damage to the brain, and (b) quantitative measurement of brain reserve capacity (the brain size, number of neurons and projections, number of synapses, etc). Brain’s functional capacity decline when neurodegenerative processes diminish brain reserve capacity below a specific limit. In contrast to brain reserve, the cognitive reserve states that the threshold for the functional decline can be changed based on experience rather than being determined by quantitative brain measurements. As a result, people with the same brain reserve capacity can have varying degrees of cognitive reserve [17], [18], [3], [14], [197], [6]. The different cognitive exposures and activities throughout the lifespan can have an impact on current brain activity and cognitive reserve. The education level, heightened intellectual ability (vocabulary or knowledge), literacy level, degree of occupational complexity, and socioeconomic position contributes to reshape the neuronal circuitries and the compensatory cognitive processes [229], [173]. The ‘JC’ used his ultimate cognitive reserve which he developed by repeating intellectually stimulating leisure activities to recall some of the memories while under similar conditions, others could not.

There is evidence suggesting that memory functions can be regained even after damage through some unknown plastic mechanisms [249], [254], [120]. However, whether all forms of memories can be compensated is still not known. Available animal models, as well as human studies, suggest that some forms of memories may be compensated [235].

Summary

A wide variety of conditions, like normal aging, psychiatric conditions, and neurological disorders, can produce memory impairment. Also, anxiety disorders like depression, bipolar mania, and schizophrenia may cause memory impairment by affecting attention, concentration, and information processing [64]. Memory is reported to be severely affected by some neurological disorders like Alzheimer’s and Parkinson’s disease. Memory disorder in Alzheimer’s disease is complex and changes with the progression of the disease. Initial symptoms of memory impairment occur in recalling recent information. In Parkinson’s disease, memory impairment is strongly correlated with motor control. Older patients also show signs of dementia. Similarly, patients suffering from Huntington’s disease show deficits in the acquisition and retrieval of memory [64], [135].

Injury to the brain has been reported to induce signs of repair and formation of new connections in neurons proximal and distal to the damaged site [139], [181], [242], [215]. Some studies have proposed that the changes occurring in compensatory mechanisms are similar to that observed during normal brain development. These compensatory changes include circuit reorganization, neurogenesis, axonal growth, and dendritic plasticity [32], [176], [42], [175]. In a primate study, lesion of the primary motor cortex associated with hands led to compensatory growth of the ventral region of the premotor cortex (PMv) in the same hemisphere. This increased extension was mediated due to the modified axonal sprouting [192], [49]. Interestingly, some compensatory brain mechanisms have also been reported in neurodegenerative diseases like Alzheimer’s disease. Studies suggest that during the early and late periods of Alzheimer’s disease, the level of proteins inhibiting the amyloid beta increase in the brain. This compensatory process acts as a defense mechanism against the aggregation and spread of amyloid-beta, which is considered the hallmark pathology of Alzheimer’s disease [21]. Although there are many unanswered questions, for example, what molecular and cellular events drive the development of compensatory circuitries, whether a greater repetitive cognitive practice simply results in increased neurotrophic factors or rehabilitation intensity is directly related to the upregulation of greater synaptic number, at what point do these processes begin? The answers to these questions still await future studies. However, overall these findings elucidate the dynamic nature of the brain and its capability to adapt to new challenges and show the path of new research to understand the mystery of our brain.

Fundings

This work was supported by DBT India, DST (CSIR) and DST (PURSE) India, UPOE-II, UGC-CAS and, UGC-Resource Networking India, ICMR India funds.

Disclosure

We disclose no conflict of interest. The funding agencies have not played any role in research design, data collection, or data analysis.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.