The C50m response: Conditioned magnetocerebral activity recorded from the human brain

Abstract

Recent advances in neuroimaging technology now permit a precise determination of the dynamics of specific neural activity underlying human associative learning. We used magnetoencephalography (MEG) to characterize the dynamics of conditioned responses (CRs) within auditory cortex during habituation, delay and trace conditioning training, and delay conditioning extinction. Conditioned stimuli (CS) were visually presented geometric figures, and unconditioned stimuli (US) were aversive noise bursts. CS+ stimuli were paired with the US on 50% of presentations: CS− stimuli were never paired with the US. Auditory cortex was activated following the paired CS+ at an average of 49–62 ms following US onset. Our data support the presence of a differential conditioned response (C50m) in auditory cortex following the unpaired CS+ at an average of 30–61 ms after US omission. The current source strength of the auditory C50m was subsequently quantified for the unpaired CS+ and CS− during training, the unpaired CS+ during extinction, and habituation. During delay and trace training, the C50m was stronger for the unpaired CS+ than for the CS−, and was also stronger for the unpaired CS+ during training compared to both habituation and extinction. This is the first description of magnetocerebral conditioning in normal human auditory cortex. The C50m activity in auditory cortex elicited by visual stimuli constitutes a direct observation of associative neural plasticity within the human auditory cortex.

Introduction

The precise dynamics of neural activity underlying normal human associative learning can now be observed due to advances in neuroimaging technology. This effort builds on previous electrophysiological studies in animal models of plastic changes in brain function observed during classical conditioning. Aversive classical conditioning serves as an excellent model for the study of associative learning in humans, as the underlying neural substrates are well characterized in many species (Carew et al., 1981a, Carew et al., 1981b, Knowlton and Thompson, 1992, Phillips and LeDoux, 1992, Solomon et al., 1986, Stanton, 2000, Supple and Leaton, 1990, Sutherland and McDonald, 1990, Walters et al., 1979, Walters et al., 1981), and conditioning paradigms can be adapted with some modifications for human neuroimaging studies (Buchel et al., 1999, Morris et al., 1998, Morris et al., 2001).

Studies of classical conditioning often involve the association of a naturally evoked autonomic response or reflex with a salient event. However, conditioned responses (CR) can also be measured directly from the brain using electrophysiology. Measures of “electrocerebral conditioning” recorded within the brain may not necessarily correlate with specific behavioral responses traditionally used to validate learning (Morrell, 1961). However, detection of electrocerebral conditioning may evidence plasticity in brain areas that are affected by conditioning, even when no overt changes in a particular behavioral response are observed.

Classical conditioning may elicit plasticity in even the earliest stages of sensory processing. Associative plasticity in sensory cortex can be studied through sensory–sensory conditioning, in which the presentation of a conditioned stimulus (CS) of one sensory modality, repeatedly paired with an unconditioned stimulus (US) of another modality, elicits a CR in sensory cortex normally activated by the US (Morrell, 1961). Sensory–sensory conditioning has been reported also in non-human invasive recording studies (Oleson et al., 1975). CRs occur in primary auditory and somatosensory cortices after repeated parings of an acoustic CS and somatosensory US. Responses in auditory and somatosensory cortices to an auditory stimulus (CS+) repeatedly paired with the US are greater than responses to another auditory stimulus (CS−) that was never paired, and than to the same CS+ prior to pairing (Dolbakyan, 1982, Oleson et al., 1975). Following pairing of a tone CS with a somatosensory US, a CR in auditory cortex occurs during US onset, which is absent prior to training or following extinction (Quirk et al., 1997).

Conditioned sensory–sensory responses have been observed in humans using electroencephalography (EEG) as alpha suppression, synchronous slow wave activity and alterations in evoked potentials (Morrell, 1961). Differential-evoked CRs are observed to CS+ versus CS− using an auditory (Hugdahl and Nordby, 1991) or somatosensory (Wong et al., 1997) US. CRs have been found 50 ms following somatosensory US omission elicited by a visual CS+, which are similar, but not identical, to unconditioned US activity (Skrandies and Jedynak, 2000). Although EEG records the dynamics of human neuronal CRs, underlying source locations remain uncharacterized.

Information about the location of CR generators has been obtained from positron emission tomography (PET) (Morris et al., 1998) and functional magnetic resonance imaging (fMRI) (Buchel et al., 1999, Morris et al., 2001). Differential auditory cortex activation has been observed following presentations of CS+ with the acoustic US omitted, versus CS−. However, direct comparison with EEG results is difficult, since PET and fMRI provide imprecise temporal information.

Magnetoencephalography (MEG) measures magnetic fields that result from neuronal current flow with comparable temporal resolution to EEG. Inversion algorithms allow for characterization of both signal dynamics and source localization (Hämäläinen et al., 1993). Prior MEG studies have demonstrated that a visual CS+ evokes conditioned activity in primary somatosensory cortex, near that elicited by the US, which is absent for a CS− and following extinction (Wik et al., 1996, Wik et al., 1997). We used MEG to further investigate the spatio-temporal properties of human associative plasticity during “magnetocerebral” conditioning. We hypothesized that following pairings of a visual CS+ with an auditory US, an unpaired CS+ would elicit CRs in auditory cortex. This would constitute a direct observation of neural plasticity within the human auditory cortex.