The Simulation of Smiles (SIMS) model holds that observing a smile leads to the extraction of meaning via different behavioral and neural processes on in the perceiver's part. In what they call their core model, Niedenthal et al. propose that the same smile is interpreted in three different ways, according to its functional meaning as enjoyment, affiliation, or dominance, by engaging different brain areas (see their Figure 4). By doing so, the SIMS model subscribes to some kind of mind-brain identity theory, establishing a one-to-one mapping of mental states to brain processes. Although such a philosophy of mind is implicit in most of modern cognitive neuroscience, SIMS is very explicit in assigning a unique, causal role to the brain processes and leaving no room for individual and situational factors which may also affect how a particular smile is interpreted. Moreover, the SIMS model appears to ignore the role of subjective and high-level conceptual processes which are crucial in emotion processing (e.g., using one's knowledge about the person who is smiling or the context in which the smile takes place). Finally, disentangling the different stages in the model is difficult, which complicates the process of deriving specific hypotheses from it and testing them empirically. Niedenthal et al. provide a fascinating synthesis of the huge literature on embodied simulation and emotion, both in psychology and the neurosciences, but distinguishing between perceptual processes, motor mimicry, and empathizing in particular cases of emotion understanding turns out to be quite a challenge. In the remainder of this commentary, we will make these arguments more specific by trying to apply the SIMS model to autism.
Autism is an early-onset neurodevelopmental disorder characterized by the co-occurrence of impairments in social reciprocity and communication, accompanied by stereotyped and rigid patterns of interests and activities. Early behavioral signals of autism, such as atypical eye contact, absence of social smiles, and absence of gaze following, are closely related to the processing of facial expressions (including smiles). At the cognitive level, children and adults with autism often have difficulties in ascribing mental states to others, referred to as mentalizing or Theory of Mind problems (e.g., Tager-Flusberg Reference Tager-Flusberg2007). This means that they experience problems in placing themselves in the mental worlds of others and in empathizing with others. Therefore, autism provides an ideal test case to evaluate the SIMS model. Applying SIMS to autism (SIMS-A) should yield additional insight or at least testable hypotheses, if the model pretends to be more than hocus-pocus with component processes and neural circuits (cf. magic words “Sim Sala Bim”).
First, linking psychological processes to neural mechanisms is definitely a way to make progress, also in research aimed at understanding the causal pathways behind a complex disorder like autism. There is a significant genetic component to autism, and some of the genes involved are clearly affecting the developing brain (e.g., Bourgeron Reference Bourgeron2009). There is a huge body of literature aimed at discovering structural and functional differences between brains of people with and without autism (e.g., Amaral et al. Reference Amaral, Schumann and Nordahl2008). However, our understanding of the vagaries and complexities of information and emotion processing in autism cannot be reduced to looking only inside the brain. If people with autism have difficulties in picking up subtle cues from faces, differentiating relevant from irrelevant stimulus changes in faces and facial expressions, integrating local details with more global context information and bottom-up perceptual information with top-down knowledge about the situation, and especially in doing all of this automatically and without painstaking effort, it is clear that an exclusive focus on neural circuits alone will not tell the whole story. At least not at the present stage of our scientific endeavor, when linking functional explanations at the psychological level to possible neural correlates is still a tricky business as far as complex, multi-factorial phenomena are concerned.
Second, autism researchers often take individual and situational factors into account when trying to explain the inconsistencies between studies finding anomalies in emotion processing in children and adults with autism and studies that do not. Clinical samples differ in factors such as age, language skills, general intelligence, subset and severity of symptoms, etc. Moreover, researchers have used a huge variety of stimuli (e.g., faces in isolation vs. embedded in complex scenes and events, static vs. dynamic facial expressions, uni- vs. multi-modal, brief vs. unlimited presentations), tasks (e.g., explicit vs. implicit, focused vs. distributed attention, sorting, matching, labeling), and measurements (e.g., performance levels, response times, eye movements). All of these factors have shown to influence the outcome of studies, but none of them is currently incorporated into the SIMS model.
Third, the SIMS model claims to be able to distinguish the role of processes like establishing eye contact, facial mimicry, simulating embodied experience, using conceptual knowledge, social inhibition, and so forth; but we are not convinced that it succeeds in doing so. For instance, can SIMS distinguish, functionally or neurally, cases of embodied emotion simulation triggered by bottom-up facial cues or by top-down knowledge? How do we differentiate the SIMS model from the many different models making use of components from the so-called mirror neuron system (e.g., Rizzolatti & Fabbri-Destro Reference Rizzolatti and Fabbri-Destro2008)? If all of these crucial yet relatively basic distinctions appear to be difficult, what about all the subtleties which appear to be involved in the literature on autism? Will it be possible to specify the SIMS model sufficiently so that testable hypotheses can be derived from it? We hope so, but there is clearly still a very long way to go.
Addressing these three general issues regarding the SIMS model and their specific elaboration regarding autism, will clarify what kind of model SIMS is, and what it can and cannot offer.
The Simulation of Smiles (SIMS) model holds that observing a smile leads to the extraction of meaning via different behavioral and neural processes on in the perceiver's part. In what they call their core model, Niedenthal et al. propose that the same smile is interpreted in three different ways, according to its functional meaning as enjoyment, affiliation, or dominance, by engaging different brain areas (see their Figure 4). By doing so, the SIMS model subscribes to some kind of mind-brain identity theory, establishing a one-to-one mapping of mental states to brain processes. Although such a philosophy of mind is implicit in most of modern cognitive neuroscience, SIMS is very explicit in assigning a unique, causal role to the brain processes and leaving no room for individual and situational factors which may also affect how a particular smile is interpreted. Moreover, the SIMS model appears to ignore the role of subjective and high-level conceptual processes which are crucial in emotion processing (e.g., using one's knowledge about the person who is smiling or the context in which the smile takes place). Finally, disentangling the different stages in the model is difficult, which complicates the process of deriving specific hypotheses from it and testing them empirically. Niedenthal et al. provide a fascinating synthesis of the huge literature on embodied simulation and emotion, both in psychology and the neurosciences, but distinguishing between perceptual processes, motor mimicry, and empathizing in particular cases of emotion understanding turns out to be quite a challenge. In the remainder of this commentary, we will make these arguments more specific by trying to apply the SIMS model to autism.
Autism is an early-onset neurodevelopmental disorder characterized by the co-occurrence of impairments in social reciprocity and communication, accompanied by stereotyped and rigid patterns of interests and activities. Early behavioral signals of autism, such as atypical eye contact, absence of social smiles, and absence of gaze following, are closely related to the processing of facial expressions (including smiles). At the cognitive level, children and adults with autism often have difficulties in ascribing mental states to others, referred to as mentalizing or Theory of Mind problems (e.g., Tager-Flusberg Reference Tager-Flusberg2007). This means that they experience problems in placing themselves in the mental worlds of others and in empathizing with others. Therefore, autism provides an ideal test case to evaluate the SIMS model. Applying SIMS to autism (SIMS-A) should yield additional insight or at least testable hypotheses, if the model pretends to be more than hocus-pocus with component processes and neural circuits (cf. magic words “Sim Sala Bim”).
First, linking psychological processes to neural mechanisms is definitely a way to make progress, also in research aimed at understanding the causal pathways behind a complex disorder like autism. There is a significant genetic component to autism, and some of the genes involved are clearly affecting the developing brain (e.g., Bourgeron Reference Bourgeron2009). There is a huge body of literature aimed at discovering structural and functional differences between brains of people with and without autism (e.g., Amaral et al. Reference Amaral, Schumann and Nordahl2008). However, our understanding of the vagaries and complexities of information and emotion processing in autism cannot be reduced to looking only inside the brain. If people with autism have difficulties in picking up subtle cues from faces, differentiating relevant from irrelevant stimulus changes in faces and facial expressions, integrating local details with more global context information and bottom-up perceptual information with top-down knowledge about the situation, and especially in doing all of this automatically and without painstaking effort, it is clear that an exclusive focus on neural circuits alone will not tell the whole story. At least not at the present stage of our scientific endeavor, when linking functional explanations at the psychological level to possible neural correlates is still a tricky business as far as complex, multi-factorial phenomena are concerned.
Second, autism researchers often take individual and situational factors into account when trying to explain the inconsistencies between studies finding anomalies in emotion processing in children and adults with autism and studies that do not. Clinical samples differ in factors such as age, language skills, general intelligence, subset and severity of symptoms, etc. Moreover, researchers have used a huge variety of stimuli (e.g., faces in isolation vs. embedded in complex scenes and events, static vs. dynamic facial expressions, uni- vs. multi-modal, brief vs. unlimited presentations), tasks (e.g., explicit vs. implicit, focused vs. distributed attention, sorting, matching, labeling), and measurements (e.g., performance levels, response times, eye movements). All of these factors have shown to influence the outcome of studies, but none of them is currently incorporated into the SIMS model.
Third, the SIMS model claims to be able to distinguish the role of processes like establishing eye contact, facial mimicry, simulating embodied experience, using conceptual knowledge, social inhibition, and so forth; but we are not convinced that it succeeds in doing so. For instance, can SIMS distinguish, functionally or neurally, cases of embodied emotion simulation triggered by bottom-up facial cues or by top-down knowledge? How do we differentiate the SIMS model from the many different models making use of components from the so-called mirror neuron system (e.g., Rizzolatti & Fabbri-Destro Reference Rizzolatti and Fabbri-Destro2008)? If all of these crucial yet relatively basic distinctions appear to be difficult, what about all the subtleties which appear to be involved in the literature on autism? Will it be possible to specify the SIMS model sufficiently so that testable hypotheses can be derived from it? We hope so, but there is clearly still a very long way to go.
Addressing these three general issues regarding the SIMS model and their specific elaboration regarding autism, will clarify what kind of model SIMS is, and what it can and cannot offer.