In search of a recognition memory engram

Abstract

A large body of data from human and animal studies using psychological, recording, imaging, and lesion techniques indicates that recognition memory involves at least two separable processes: familiarity discrimination and recollection. Familiarity discrimination for individual visual stimuli seems to be effected by a system centred on the perirhinal cortex of the temporal lobe. The fundamental change that encodes prior occurrence within the perirhinal cortex is a reduction in the responses of neurones when a stimulus is repeated. Neuronal network modelling indicates that a system based on such a change in responsiveness is potentially highly efficient in information theoretic terms. A review is given of findings indicating that perirhinal cortex acts as a storage site for recognition memory of objects and that such storage depends upon processes producing synaptic weakening.

Keywords: Familiarity; Imprinting; LTD; LTP; Perirhinal cortex.

Fig. 1

Schematic diagram of the paired-viewing apparatus. When the rat's snout is engaged in the observing hole in the Perspex screen, the right eye can see only the right monitor and the left eye can see only the left monitor. Stimuli are presented on both monitors simultaneously. Juice is delivered towards the end of picture presentation through a tube that the rat can just reach to lick when its snout is in the observing hole.

Fig. 2

Rodent object recognition memory task. A rat is exposed to and allowed to explore two copies of an object during the sample phase. After a delay interval, the rat is allowed to explore a third copy of the object already explored in the sample phase and a novel object. Normal rats spend more time exploring the novel than the familiar object in the test phase. Infusions of drugs may be made prior to the sample phase to interfere with acquisition and early consolidation, at various times after the sample phase during the delay interval to interfere with consolidation and storage mechanisms, or prior to the test phase to interfere with retrieval mechanisms.

Fig. 3

Schematic representation of memory decays (‘forgetting curves’) produced by selective cholinergic and glutamatergic antagonists. The curves are based on forgetting curves derived from population measures of monkey perirhinal neuronal responses to novel and familiar stimuli after different delays for novelty (N) and familiarity (F) neuronal types (Xiang and Brown, 1998). If kainate (KAR) and muscarinic receptor antagonists target ‘novelty’ (fast synaptic change) neurons, while NMDA and nicotinic receptor antagonists target ‘familiarity’ (slow synaptic change) neurons, then the different forgetting curves provide potential explanation for the different amnesic effects observed: NMDA or muscarinic antagonism results in short-term memory followed by forgetting at longer intervals, whereas kainate (KAR) or muscarinic antagonism produces short-term forgetting followed by long-term remembrance.

Fig. 4

Novelty detection latency in monkey temporal cortex. Top left panel: Responses of a neuron to presentations of novel and familiar stimuli. Peristimulus histograms show the average firing rate for novel and for familiar stimuli. Dots beneath each histogram show the times of occurrence of individual action potentials on each trial on which a single novel or familiar stimulus was shown. Bottom left panel: Cumulative action potential count after stimulus onset for the novel and familiar trials. A statistically significant difference was established by the 60–90 ms time bin. Top right panel: Population average of such individual neuronal cumulative action potential counts for neurons whose responses change with stimulus familiarity. Novel and familiar population responses first differ significantly in the 60–90 ms time bin (Xiang and Brown, 1998) (Left two panels adapted with permission from Brown and Bashir, 2002). Bottom right panel: Human MEG signals for novel and familiar stimuli. There is a larger signal for novel than familiar stimuli in the 85–115 ms time bin when subjects’ responses earned reward (the asterisk denotes a significant difference). The differential latency is closely similar to that in the monkey when account is taken of the difference in monkey and human brain sizes.

Fig. 5

Phosphorylation of CamKII and familiarity discrimination. Top panel: Effect of perirhinal infusions of vehicle or the CAMKII inhibitor AIP (autocamtide-2-related inhibitory peptide) at differing times relative to acquisition on object recognition memory after a 24 h delay in the rat. Discrimination ratios (DR) were used as an index of memory performance (a DR of zero indicates no preference). *Time-points at which there was a significant effect of treatment. Amnesia followed infusions 20–60 min after acquisition. Lower two panels: Normalised counts of pCAMKIIa-stained neurons at different times after the paired viewing of novel and familiar stimuli for (middle panel) perirhinal region (PRH), (bottom panel) area Te2, *Significant differences between novel and familiar counts were found 70 min after viewing in both regions.

Fig. 6

LTD pathways in perirhinal cortex. (1) Activation of glutamatergic and cholinergic afferents in perirhinal cortex, typically at 1–5 Hz, leads to release of neurotransmitter and activation of post-synaptic receptors. (2) Kainate receptor (KAR) activation is required for familiarity discrimination at short (≤20 min) but not longer delays. Although there are no currently known roles for kainate receptors in perirhinal plasticity, it is known that inhibition of AP2-dependent AMPA receptor endocytosis by pepD849-Q853 also impairs object recognition memory at short delays, thus suggesting that KARs are also involved in synaptic weakening processes. (3) Cholinergic modulation of perirhinal cortex is implicated in both learning and protein synthesis-dependent LTD. Muscarinic acetylcholine receptor 1 (mAChR1) activation leads to release of calcium from intracellular stores and subsequent activation of calcium-sensitive kinases, and additionally stimulates extracellular-signal related kinases (ERK) leading to phosphorylation of CREB and production of Fos protein. Muscarinic receptor activation also activates nitric oxide synthase (NOS), producing nitric oxide (NO) which can act as a retrograde signalling molecule, activating soluble guanylate cyclase which attenuates glutamate release. Block of mAChR1s during acquisition impairs object recognition memory at delays of up to 6 h, whilst inhibition of NOS impairs memory at a delay of 24 h, suggesting mechanisms other than mAChR1 may also stimulate NO production in perirhinal cortex. (4) Activation of L-type voltage gated calcium channels (VGCCs), mGluRs and GluN2B-containing NMDA receptors are all required for object recognition at a 24 h delay. Activation of these proteins leads to increases in intracellular calcium concentration and calcium–calmodulin dependent kinase (CamK) activation which is thought to phosphorylate AMPA receptors and facilitate their endocytosis. (5) CREB phosphorylation is required for object recognition memory at a 24 h delay and is increased by mGluR, mAChR1 and calcium–calmodulin dependent kinase (CamK) activation. Phosphorylated CREB stimulates transcription and is known to lead to production of Fos protein. Although these events are required for object recognition memory and LTD, it is currently unclear how these processes lead to synaptic weakening. It is however known that endocytosis of AMPA receptors by clathrin adaptor protein AP2 is required for both LTD and object recognition memory at delays of 5 min or 24 h.