Perception and the Medial Temporal Lobe: Evaluating the Current Evidence Wendy A. Suzuki NYU

Outline Anatomy of the medial temporal lobe (MTL) Medial temporal lobe memory system (MTLMS) hypothesis – Evidence towards MTLMS from humans and non-human primates Perceptual-mnemonic (PM) hypothesis – Evidence against PM from humans, non-human primates and rats

Human Medial Temporal Lobe (MTL) Hippocampus Entorhinal cortex Perirhinal cortex Parahippocampal cortex Kandel et al., 2000

Animal MTL Anatomy

Medial Temporal Lobe Memory System (MTLMS) Traditional view: – MTL is involved in the ability to learn and retain declarative memory – Declarative memory involves information for events, facts and relationships (Eichenbaum and Cohen, 2001; Squire and Zola-Morgan, 1991) – Procedural learning, perceptual priming, and emotional memory are spared

Perceptual-Mnemonic Hypothesis Novel view: – MTL is involved in declarative memory AND perception (Bussey and Saksida, 2007; Lee et al., 2005; Murray and Wise, 2004) – Specifically, the perirhinal cortex – High feature ambiguity (overlapping features)

Perception in the Lateral Temporal Lobe Ventral Stream (Vision for Perception) Dorsal Stream (Vision for Action) Milner and Goodale, 1995

Wendy Suzuki: Argument I argue that little or no convincing evidence exists in favour of the role of the MTL in perception

MTLMS: Evidence from Human Amnesic Patients Patient H.M. ( Henry Gustav Molaison) – Mostly anterograde amnesia, but also temporally graded retrograde amnesia – Bilateral removal of hippocampus (2/3), amygdala, parahippocampal gyrus, perirhinal and entorhinal cortices – Declarative memory gone, perception spared

MTLMS: Evidence from Non-human Primates Induce lesions – Large MTL lesions modeled from H.M. ( Mishkin, 1978; Zola- Morgan and Squire, 1985) – Selective MTL lesions – perirhinal and entorhinal, just perirhinal (Meunier et al., 1993), or hippocampus (Alvarez et al., 1995; etc) All involved in declarative/relational memory with online perceptual processing spared – Can still process features (colours/shapes) – Can differentiate between stimuli

MTLMS: Evidence from Non-human Primates MTL receives inputs from many high-order sensory and association areas MTL might be sight for integration of perceptual stimuli (Suzuki and Amaral, 1994)

Delayed Nonmatching-to-Sample (DNMS) Task Zola-Morgan et al., 1989 ( Mishkin, 1978)

Perirhinal Lesions: No Perceptual Deficits in Monkeys Tested for perceptual generalization (Hampton and Murray, 2002) – Modified stimuli (rotated, enlarged, shrunken, or masked) – Criticism for similarity to Shepard–Metzler DNMS task (Buffalo et al., 2000) – Two superimposed objects (high overlap) – Short delay (0.5s)

Monkey Perirhinal Lesions: Evidence Against the PM Hypothesis Studies claiming MTL involvement in perception have: – Confounded impairment in learning/memory – Required working memory in perception tasks – Extent of the lesion not identified

Monkey Perirhinal Lesions: Evidence Against the PM Hypothesis Bilateral rhinal ablation (Eacott et al., 1994) – Reported perceptual impairment in 0s delay and simultaneous matching conditions – Large stimulus sets – Averaged together – Later reported that some samples were left out – When left out, no difference! Buckely and Gaffan, 1998

Monkey Perirhinal Lesions: Evidence Against the PM Hypothesis Concept of visual discrimination learning –Can you test perception in animals without engaging memory? –Learning over multiple trials –Studies reported spared ‘elemental’ discriminations, but impaired ‘object’ discrimination (Buckley and Gaffan, 1998; Bussey et al., 2002, 2003) Buckely and Gaffan, 1998

Oddity Discrimination Task Developed in attempt to tease out perception from memory – No difference in perirhinal lesioned monkeys during easy and moderate discrimination (colours, shapes) – Problems during complex object discrimination – Perhaps implies impairment in visual associative learning to associate multiple face views as one (Messinger et al., 2001; etc) Buckley et al., 2001

Oddity Discrimination Task Find the scene outlier – Only 10 scenes – Could use long-term memory to recall the identities of each scene. Buckley et al., 2001

Morphing Features Different stages of morphing flowers – Each one is morphed into previous one (cumulative) “High feature overlap” –High feature ambiguity must be held in working memory (Bussey, 2006) Learning over multiple trials (Bussey et al., 2002) – Must retain object/spatial features in working memory Buckley et al., 2001

Rat Perirhinal Lesions: Evidence Against the PM Hypothesis Bilateral ablation (Bartko et al., 2007) Simultaneous oddity discrimination task – Impaired discrimination with high feature ambiguity – PROBLEM: large objects used relative to rats

Summary Thus Far: MTLMS MTL is involved in only declarative memory – No perceptual/procedural deficits Studies that support PM hypothesis are unconvincing – Do not adequately isolate perceptual demands from declarative/relational learning or memory Results are inconsistent

Human MTL Lesions Major advantages of working with humans: – Verbal/written instructions – Probe to ensure instructions are followed – Follow-up of subject’s opinions However, many contradictory outcomes in the literature exist

Human MTL Lesions: Evidence in Favour of MTLMS Hypothesis Large bilateral MTL/hippocampal lesions (Stark and Squire, 2000) Performed same oddity discrimination task as Buckley et al., 2001 – No perceptual impairment – Maybe prefrontal inputs associated with rules of task?

Human MTL Lesions: Evidence Against the PM Hypothesis Lee et al., 2005 found impaired discrimination in complex face and scene oddity tasks – Larger stimulus sets – Simultaneous presentation of target and test item (faces/VR rooms) – Added trial-unique discrimination task

Reasons for Discrepancies Different testing procedures in different labs? Controls performed equally as well in both studies at all five difficulties and in all three tasks Reproduced lack of perceptual impairment for trial-unique discrimination Differences in extent/location of lesion? Some subjects also had anterior temporal, anterior insula, fusiform gyri lesions, and even lateral TE damage – TE involved in high-order visual perception (Squire, personal communication)

Reasons for Discrepancies (cont) Differences in extent/location of lesion (cont) – Lesion loci were estimated by visual inspection (4/5 rating scale) (Lee et al., 2005) – Single slice through each area (biased?) – Authors mention validation to volumetric analysis (Galton et al., 2001), but provide no details of validation Squire’s group performed detailed volumetric analysis throughout MTL and LTL Recently, MTL has been validated (Barense et al., 2007), but LTL has not

Reasons for Discrepancies (cont) Lee et al. (2005) argue that MTL group has minimal TE damage since they can discriminate colours and moderately complex objects – TE active during high overlap, not colours (Buckley et al, 1997) – TE neurons respond to faces (Alfred et al., 2005; etc) – TE neurons may reflect upstream input into perirhinal cortex

TE vs Perirhinal Neurons TE neurons respond to brightness of cue (Liu and Richmond, 2000) Perirhinal neurons respond to visual-reward associative information of cue -Trial schedule independent of cue TE = perception Perirhinal = link between perceptual information and memory

Conclusions MTLMS hypothesis states MTL is involved in declarative/relational learning and memory with little or no contribution to perception – 60 years old Animal models lack ability to differentiate perception from memory (strategies) Human MTL lesions are inconsistent (extent and location)

Future Directions Further pursue findings with rats (Bartko et al., 2007) – Define semantics – Tease out perceptual and working memory differences (possible continuum) – May be due to axonal connections between MTL and LTL – Add muscimol to perirhinal cortex and record from TE (in monkeys) Look for compensatory activity in humans – fMRI or connectivity analysis (fcMRI, DTI, DSI, Granger causality)

Thank you ;-)