Recognition Memory

This page discusses the research carried out here at the MRC Centre into visual recognition memory. Follow the links for information on recently published work and and important background concepts.
Visual recognition memory involves a reduction in neuronal activity
What is the mechanism of neuronal decrement?
Recognition memory involves the interaction of multiple brain regions
Multiple neurotransmitter systems are involved in recognition memory

Visual recognition memory involves a reduction in neuronal activity

The neurons that perform the basic tasks of visual recognition, such as recognising individual items in the visual field, reside in an area of the brain called the perirhinal cortex. This cortical region is part of the temporal lobe, adjacent to the hippocampus. Neurons in this area respond to novel items, such as an image that has not been seen before, but respond much less to familiar items. This reduction (neuronal decrement) in activity can occur very rapidly, for instance with an image that has seen just once, and can last for an extended period of time. Neurons that respond in this way are called novelty neurons. They signal that something has not been seen before.

Other related neurons signal that something is familiar (familiarity neurons); they lose their responsiveness after an image is seen number of times and do not respond to it after that. Yet more neurons signal that something has been seen recently (recency neurons); they respond the first time something is seen and then show reduced activity the next time it is seen. However, the responsiveness of these neurons to the stimulus recovers quickly. Thus if these neurons respond, the stimulus cannot have been seen recently.

What is the mechanism of neuronal decrement?

An important part of the research into recognition memory systems that is currently being pursued at the MRC Centre revolves around the investigation of the molecular mechanisms that are responsible for the reduction in synaptic activity. Such a mechanism should be necessary and sufficient to be able to account for all the features of recognition memory. The leading candidate for such a molecular explanation of this form of memory is long-term depression (LTD), a form of synaptic plasticity that results in a long lasting reduction in synaptic activity. Not only can LTD provide a sound mechanistic solution to neuronal decrement, the regulation of input specificity means that, under normal conditions, only single synapses undergo modification from each incoming signal. Thus, one novel stimulus may modify one set of synapses while a second stimulus modifies another set. Given the number of synapses typically present on a neuron (1,000 or more), this gives each neuron a large number of different stimuli that can be treated as novel/familiar. Indeed, computer modelling suggests that the human perirhinal cortex has a capacity of about 1,000,000,000 items that can be recognised as having been seen before.

Does LTD underlie recognition memory?

Experimental evidence for LTD being the molecular mechanism that underlies recognition memory comes from a number of sources. Most recently, a study carried out here at the MRC Centre used viral transduction of competing peptides to block AP2, a critical protein in the clathrin-mediated internalisation of AMPA receptors, a process known to be fundamental to the expression of LTD ( Griffiths et al 2008). It was shown that, not only did the competing peptide block LTD in in vitro slice preparations, but that it also disrupted spontaneous object recognition behaviour and LTD could not be induced in slices taken from animals transduced with this peptide.

Further support to the hypothesis that LTD underlies recognition memory comes from earlier work in which scopolamine (a cholinergic antagonist routinely used as a pre-operative sedative) and lorazepam (an anxiolytic that produces amnesia in high doses) not only prevented the recognition of novelty, but also blocked LTD (lorazepam additionally blocked LTP) and reduced the production of fos, (the product of the immediate early gene c-fos, the activity of which is used as a marker neuronal activity) in slices of perirhinal cortex (Warburton et al 2003; Wan et al 2004).

Thus, in these two examples, a single form of treatment has consistent effects across multiple levels of systems activity. It is in this way, in combination with computer modelling, that the molecular and cellular mechanisms that underlie our memory systems can be investigated.

Is LTD alone responsible for recognition memory?

LTD may not be the only mechanism repsonsible for the synaptic changes that must underlie recognoition memory. For instance, blockade of GluN2A or GluN2B receptor subunits alone do not impair memory function in the sponaneous object recognition test, while blockade of both together does. These two NMDA receptor subunits have been shown to be involved in two different forms of decremental plasticity - depotentiation and LTD, repsectively. Thus NVP-AAM007 (a GluN2A antagonist) blocks depotentiation (and LTP) but not LTD, while Ro25-6981 (a GluN2B antagonist) blocks LTD but not depotentiation ( Massey et al, 2004). Thus both depotentiation and LTD can support recognition memory, at least in the longer term. Further mechanisms may also be at work in short-term memory, as this seems to be dependent on both cholinergic muscarinic receptors and kainate receptors. Thus, there may be convergent cell signalling mechanisms that feed into either two further forms of LTD or that converge on one. It should also be remembered that there are many different molecular mechanisms that result in LTD.

(Back to top)

Recognition memory involves the interaction of multiple brain regions

Visual recognition memory requires a number of judgements other than those based on novelty/familiarity. It also invloves the spatial arrangements of familiar objects or the temporal order in which objects have been seen. These aspects of recognition memory require the integration of multiple pieces of information. Thus, we recognise when familiar furniture has been moved in a room, which requires us to remember not only what individual items are familiar, but where they were before. Similar mechanisms exist for this aspect of recognition memory as for the recognition of individual items. However, in this case there are changes in neuronal activity in the hippocampus and an adjacent region, the postrhinal cortex (POR). The changes are also complex, in that c-fos activity is higher in response to novel arrangements in the POR and the CA1 region of the hippocampus. However, activity is higher in response to familiar arrangements of items in the dendate gyrus and subiculum (Wan et al 1999). This function of the hippocampus may be related to it's role in spatial memory.

Recent evidence suggests that the perirhinal cortex (PRH) is also involved in object arrangement recognition. Using a four object variant of the spontaneous object recognition task, disruption of perirhinal cortex function results in an impairment in the ability of an animal to tell when the positions of two objects have been swapped (object-in-place task: Barker et al 2007). However, such disruption has no effect on the behaviour of an animal when one of a pair of identical items is moved to a new position (object location task). This suggests that the perirhinal cortex not only signals that two non-identical objects are familiar, but their spatial relationship to each other, though not their absolute position.

Connections between different brain regions involved in visual recognition memoryA third brain region is involved with recognition memory - the medial prefrontal cortex (mPFC). Work here at the MRC centre has confirmed and extended previous work that demonstrated that the mPFC is involved in recency discrimination but not in object location or novelty/familiarity discrimination (Barker et al 2007). Furthermore, it was shown that disruption of mPFC function leads to an impairment in the object-in-place task, suggesting a role for the mPFC in detecting the spatial relationships between objects.

When multiple brain regions have been identified as being involved in the same process, it would seem likely that they operate as a neural network with information being passed between them. Well established anatomical evidence shows that projections connect the hippocampus and the mPFC while there are also anatomical connections between the mPFC and the perirhinal cortex. The perirhinal cortex is also connected to the hippocampus via the entorhinal cortex (see Brown & Aggleton 2001 for review). It is well known that disruption of the fornix (the fibre bundle that forms a major output of the hippocampus to other parts of the limbic system and to the mPFC) or the PRH-POR connection impairs behaviour in the object-in-place task (e.g.Bussey et al 2000). We have also recently shown as part of the above study that disruption of the connection between PRH and mPFC and also results in impairment in this task. So it would appear that a functioning hippocampus, perirhinal and medial prefrontal cortices are necessary for this form of spatial discrimination along with their interconnections.

(Back to top)

Multiple neurotransmitter systems are involved in recognition memory

Effects of scopolamine on memory aquisition

Blockade of cholinergic transmission with scopolamine prior to memory acquisition impairs performance in spontaneous object recognition tasks. Mounted on the internet with permission of Warburton et al, 2003 Neuron 6; 987-996 © Cell Press Ltd

Disruption of familiarity discrimination has shown that there are multiple neurotransmitter and receptor systems involved in recognition memory. For instance, blockade of perirhinal muscarinic receptors with scopolamine during the acquisition phase of a spontaneous object recognition task, i.e. when the animal is familiarising itself with an object, reduces it's ability to discriminate between the object it's already seen and a new one. It will spend about an equal period of time investigating both novel and familiar objects. However, if scopolamine is administered after the acquisition phase, there is no effect on familarity discrimination (Warburton et al 2003). Thus scoploamine does not make the animal forget but prevents the proper formation and/or consolidation of the memory. Similar results have been seen using lorazepam (Wan et al 2004), a benzodiazapine used as an axiolytic that acts on GABAA receptors, thus implicatingGABAergic transmission in recognition memory.

Scopolamine has also been shown to disrupt memory aquisition in the object-in-place task (Barker and Warburton, 2008) when applied bilaterally to either PRH or mPFC regions. Combined unilateral blockade of muscarinic receptors in these two regions also disrupted this form of memory. Thus, we have shown not only that these regions act together to form a memory network, but that they depend on at least cholinergic neurotransmission.

Effect of AP5 and UBP302 on memory aquisitionLong-and short-term recognition memory rely on different glutmatergic transmission mechanisms. Mounted on the internet with permission of Barker et al, 2006 J. Neurosci. 26; 3561-3566 © Society for Neuroscience
The situation with glutamatergic transmission is more complex. Recent findings ( Barker et al 2006a) have demonstrated that at least two, and possibly three or more, independent mechanisms with different temporal profiles exist within the perirhinal cortex. One for short-term memory formation relies on kainate receptors, whereas long-term memory formation relies on NMDA receptors, acting through two separate processes. Blockade of NMDA receptors during the acquisition phase of the spontaneous object discrimination task using the antagonist D-AP5 results in an impairment of object recognition when the animals are tested again 24 hours but if tested 20 minutes later. Similar results are seen when antagonists of GluN2A and GluN2B subunits are used together, but not when used alone. Thus, two mechanisms that are dependent on GluN2A and GluN2B subunits can support long-term recognition memory. Conversly, blockade of GluK1-containing kainate receptors with UBP302 (a selective GluK1 subunit antagonist developed by David Jane) results in impairment of object recognition after 20 minutes but not 24 hours ( Barker et al 2006a). It is interesting to note that there is evidence that the GluN2A and GluN2B subunits of the NMDA receptor may be localised in different synaptic compartments and be involved in different forms of LTD in the adult rat perirhinal cortex (Massey et al 2004).

In addition, a further mechanism for long term memory requires the synergistic activation of group I and group II metabotropic glutamate receptors. Blockade of both mGlu5 receptors with MPEP and group II receptors with LY341495 impaired familiarity discrimination at a 24hr but not a 15 min delay between training and trial. Treatment with either antagonist alone had no effect (Barker et al 2006b). Whether the effects of blocking NMDA or mGlu receptors affects the same molecular process is not known, although synergy between these receptors in the generation of LTD has been reported previously (Cho et al 2000). It may be that activation of all three receptor systems is required for the aquisition of long-term object recognition.

(Back to top)

Note: Some of the pages on this site contain Flash documents. To view these documents, you will need a small, free program Adobe Flash Player