Affect Recognition Research Articles

Interplay among positive and negative symptoms, neurocognition, social cognition, and functional outcome in clinically stable patients with schizophrenia: a network analysis.

Background: Schizophrenia has a broad range of interrelated symptoms and impairment in functioning. The objective of the study was to explore the interplay between positive symptoms, negative symptoms, neurocognition, social cognition and functional outcome in patients with schizophrenia using network analysis. Methods: Participants were 64 clinically stable patients with schizophrenia. Psychopathologic, neurocognition, social cognition, and functional outcome were measured using the Scale for the Assessment of Positive Symptoms, Scale for the Assessment of Negative Symptoms, Addenbrooke's Cognitive Examination III, Faces test, Reading the Mind in the Eyes test,and Personal Social Performance scale. Results: The network analysis suggested that functional outcome was the most central in the network followed by avolition and asociality. Functioning was directly connected to avolition, asociality, blunted affect, neurocognition and emotion recognition. The positive symptoms were the most remote and therefore the least important node. Conclusion: The high centrality of functioning suggests the need for improving of everyday life skills for patients with schizophrenia. Moreover, treatment of specific negative symptoms, neurocognition and emotion recognition could also enhance functional outcome.

User Identity Protection in Automatic Emotion Recognition through Disguised Speech

Ambient Assisted Living (AAL) technologies are being developed which could assist elderly people to live healthy and active lives. These technologies have been used to monitor people’s daily exercises, consumption of calories and sleep patterns, and to provide coaching interventions to foster positive behaviour. Speech and audio processing can be used to complement such AAL technologies to inform interventions for healthy ageing by analyzing speech data captured in the user’s home. However, collection of data in home settings presents challenges. One of the most pressing challenges concerns how to manage privacy and data protection. To address this issue, we proposed a low cost system for recording disguised speech signals which can protect user identity by using pitch shifting. The disguised speech so recorded can then be used for training machine learning models for affective behaviour monitoring. Affective behaviour could provide an indicator of the onset of mental health issues such as depression and cognitive impairment, and help develop clinical tools for automatically detecting and monitoring disease progression. In this article, acoustic features extracted from the non-disguised and disguised speech are evaluated in an affect recognition task using six different machine learning classification methods. The results of transfer learning from non-disguised to disguised speech are also demonstrated. We have identified sets of acoustic features which are not affected by the pitch shifting algorithm and also evaluated them in affect recognition. We found that, while the non-disguised speech signal gives the best Unweighted Average Recall (UAR) of 80.01%, the disguised speech signal only causes a slight degradation of performance, reaching 76.29%. The transfer learning from non-disguised to disguised speech results in a reduction of UAR (65.13%). However, feature selection improves the UAR (68.32%). This approach forms part of a large project which includes health and wellbeing monitoring and coaching.

COVID-19: Affect recognition through voice analysis during the winter lockdown in Scotland.

The COVID-19 pandemic has led to unprecedented restrictions in people's lifestyle which have affected their psychological wellbeing. In this context, this paper investigates the use of social signal processing techniques for remote assessment of emotions. It presents a machine learning method for affect recognition applied to recordings taken during the COVID-19 winter lockdown in Scotland (UK). This method is exclusively based on acoustic features extracted from voice recordings collected through home and mobile devices (i.e. phones, tablets), thus providing insight into the feasibility of monitoring people's psychological wellbeing remotely, automatically and at scale. The proposed model is able to predict affect with a concordance correlation coefficient of 0.4230 (using Random Forest) and 0.3354 (using Decision Trees) for arousal and valence respectively.Clinical relevance- In 2018/2019, 12% and 14% of Scottish adults reported depression and anxiety symptoms. Remote emotion recognition through home devices would support the detection of these difficulties, which are often underdiagnosed and, if untreated, may lead to temporal or chronic disability.

Working Memory for Emotions in Adolescents and Young Adults with Traumatic Brain Injury.

This preliminary study investigated the interaction between working memory and social cognition in adolescents and young adults with traumatic brain injury (TBI). It was hypothesized that participants with or without TBI would better recognize social information when working memory or social cognitive load was low, and that adolescents and young adults with TBI would be more affected by increased cognitive demand than their uninjured peers. Eight adolescents and young adults with complicated mild-severe TBI (aged 14-22 years) and eight age- and sex-matched typically developing (TD) adolescents completed computer-based n-back tasks requiring recognition of either face identity or facial affect, with 0-back, 1-, and 2-back conditions. The TBI group had lower scores overall than the TD group, and scores for both groups were lower for affect recognition than identity recognition. Scores for both groups were lower in conditions with a higher working memory load. There was a significant group-by-working memory interaction, with larger group differences in high-working memory conditions. Study results and their potential implications for social outcomes are discussed.

Human-Guided Modality Informativeness for Affective States.

This paper studies the hypothesis that not all modalities are always needed to predict affective states. We explore this hypothesis in the context of recognizing three affective states that have shown a relation to a future onset of depression: positive, aggressive, and dysphoric. In particular, we investigate three important modalities for face-to-face conversations: vision, language, and acoustic modality. We first perform a human study to better understand which subset of modalities people find informative, when recognizing three affective states. As a second contribution, we explore how these human annotations can guide automatic affect recognition systems to be more interpretable while not degrading their predictive performance. Our studies show that humans can reliably annotate modality informativeness. Further, we observe that guided models significantly improve interpretability, i.e., they attend to modalities similarly to how humans rate the modality informativeness, while at the same time showing a slight increase in predictive performance.

The Impact of Sleep Deprivation on Work Performance towards Night-shift Healthcare Workers: An Evidence-based Case Report

Background. Poor sleep and excessive fatigue among workers can reduce well-being and physical fitness. However, not many studies have mentioned how sleeping deprivation among night-shift healthcare workers impacted their work performance in multiple aspects.
 Method. We conducted an evidence-based case report (EBCR) of a night shift nurse who was worried about the impact of her sleep deprivation on her work performance in the future due to prior history of needle-stick injuries. We aimed to determine whether sleeping deprivation caused by regular night shifts leads to decreased work performance among night-shift healthcare workers by formulating a clinical question. Evidence was searched systematically using five major journal databases (Proquest, EBSCO-Host, PubMed, ScienceDirect, and Cochrane) and was assessed thoroughly using inclusion and exclusion criteria.
 Results. Eleven eligible studies were obtained with a medium level of evidence (III-IV), three systematic reviews with meta-analyses (SR-MA), three SR without MA, and five observational studies. All of them were analyzed and critically appraised using Oxford Evidence-Based Medicine and Joanna Briggs Institute tools. We found that reduced quantity and quality of sleep impacted all dimensions of work performance among healthcare professionals, be it in task performance (e.g., skill proficiency), contextual performance (e.g., communication skill and mental health issues), and patient and health worker safety (accident and medication error). It could also encourage counterproductive work behavior, such as absenteeism. Furthermore, sleep deprivation changes circadian rhythms, causing decreased information processing and affective recognition functions in some vital brain areas, ultimately affecting several work dimensions.
 Conclusion. In conclusion, stakeholders need to adjust proper shift scheduling for health care workers, practice sleep hygiene, maintain physical fitness, and consume nutritional food, positively correlated to health and productivity.

Multi-model hybrid ensemble weighted adaptive approach with decision level fusion for personalized affect recognition based on visual cues

Multi-model hybrid ensemble weighted adaptive approach with decision level fusion for personalized affect recognition based on visual cues

Digital Feminism beyond Nativism and Empire: Affective Territories of Recognition and Competing Claims to Suffering in Iranian Women’s Campaigns

There has been a growing affective intensity on Farsi social media around Iranian women’s rights protests, particularly mobilizations against the compulsory hijab. This intensity has crystallized into a variety of emotions: the anger, joy, and defiance of Iranian women protesting against the hijab inside the country; the outrage at the injustice these actions communicate; and the anger and anxiety of Iranian critics on both the Right and the Left, among both the Iranian establishment and ordinary people. Yet oversimplified counterpositions between “global” and “local,” West and East, dominate both legacy-media and social-media narratives of women’s rights campaigning, resulting, as I show, in the circulation of binary genres of “authentic” versus “inauthentic” protest that generate anger and anxiety. In this vein, the traumatic experience of economic sanctions, as the master signifier of West-East conflict in Iranian online exchanges, often frames human rights discourses as inauthentic, in that these discourses are positioned as emanating from the West. Understanding the politics of emotion around these mobilizations allows one to attend more closely to the contentions and fissures that traverse women’s struggles within Iran and, I argue, to develop a politics of affective recognition as a basis for constructing feminist solidarities across and within borders. A focus on anger and rage, among other feelings, helps us not only to trace the binary oppositions that characterize online discourses around these protests but to capture their unstable ambivalence between authenticity and inauthenticity, inside and outside, their potential to point beyond existing boundaries and demarcations, and to permit new imaginings of territoriality.

Social skills and psychopathology are associated with autonomic function in children: a cross-sectional observational study.

In recent years, the increase of psychopathological disorders in the population has become a health emergency, leading to a great effort to understand psychological vulnerability mechanisms. In this scenario, the role of the autonomic nervous system (ANS) has become increasingly important. This study investigated the association between ANS, social skills, and psychopathological functioning in children. As an ANS status proxy, we measured heart rate variability (HRV). Infants admitted to the neonatal intensive care unit of the University Hospital of Padova because of preterm birth or neonatal hypoxic-ischemic encephalopathy were sequentially recruited from January 2011 to June 2013 and followed long-term up to school age in this cross-sectional observational study. We recorded 5 minutes of HRV immediately before measuring performance in social abilities tasks (affect recognition and theory of mind, NEPSY-II) in 50 children (mean age 7.4 ± 1.4 years) with and without risk factors for developing neuropsychiatric disorders due to pre-/perinatal insults without major sequelae. Children also completed extensive cognitive, neuropsychological, and psychosocial assessment. Parents were assessed with psychopathological interviews and a questionnaire (CBCL 6-18). Analysis in a robust Bayesian framework was used to unearth dependencies between HRV, social skills, and psychopathological functioning. Social task scores were associated with HRV components, with high frequency the most consistent. HRV bands were also associated with the psychopathological questionnaire. Only normalized HRV high frequency was able to distinguish impaired children in the affect recognition task. Our data suggest that ANS may be implicated in social cognition both in typical and atypical developmental conditions and that HRV has cross-disease sensitivity. We suggest that HRV parameters may reflect a neurobiological vulnerability to psychopathology. The study was approved by the Ethics Committee of the University Hospital of Padova (Comitato Etico per la Sperimentazione, Azienda Opedaliera di Padova, approval No. 1693P).

A-127 The Interplay of Social Perception and Psychiatric Symptoms in a Case of Traumatic Brain Injury

Abstract Objective Acquired impairment in social perception (SP) is relatively common following moderate to severe traumatic brain injury (TBI), with up to 39% of these individuals experiencing deficient facial affect recognition. Formal assessment of SP is not common practice, despite evidence of this as a predictor of social functioning. We present the case of a 35-year-old, Caucasian male with new onset paranoid/delusional (P/D) thought processes in the context of prior TBI. Method The patient’s TBI (unknown severity) occurred 17 years prior and resulted in prolonged (>1 year) hospitalization, followed by intensive rehabilitation services. He reported residual, stable difficulties with attention, processing speed, executive functioning, and memory. Post-TBI, he was living independently with few external supports. Following a traumatic encounter with a family member (nine months prior), the patient developed P/D thought processes and a marked decline in functioning, resulting in three psychiatric hospitalizations. Brain MRI (Figure 1) revealed encephalomalacia and gliosis in the bilateral inferior frontal lobes anteriorly, high frontoparietal lobes, and bilateral (L > R) temporal lobes. Neuropsychology was consulted to clarify diagnosis/etiology and assess functional status. Results (Table 1). Evaluation results revealed non-amnestic memory impairment and variable language and processing speed. Performances on measures of SP revealed a notable weakness in the context of current intellectual performances. Conclusions This case highlights the importance of formally assessing SP in cases of TBI. Underlying SP difficulties in the context of a traumatic event may have increased vulnerability to psychiatric decompensation. Evaluation of SP allowed for enhanced understanding of the interplay between cognitive and psychiatric symptoms.

Automatic subject-specific spatiotemporal feature selection for subject-independent affective BCI.

The dimensionality of the spatially distributed channels and the temporal resolution of electroencephalogram (EEG) based brain-computer interfaces (BCI) undermine emotion recognition models. Thus, prior to modeling such data, as the final stage of the learning pipeline, adequate preprocessing, transforming, and extracting temporal (i.e., time-series signals) and spatial (i.e., electrode channels) features are essential phases to recognize underlying human emotions. Conventionally, inter-subject variations are dealt with by avoiding the sources of variation (e.g., outliers) or turning the problem into a subject-deponent. We address this issue by preserving and learning from individual particularities in response to affective stimuli. This paper investigates and proposes a subject-independent emotion recognition framework that mitigates the subject-to-subject variability in such systems. Using an unsupervised feature selection algorithm, we reduce the feature space that is extracted from time-series signals. For the spatial features, we propose a subject-specific unsupervised learning algorithm that learns from inter-channel co-activation online. We tested this framework on real EEG benchmarks, namely DEAP, MAHNOB-HCI, and DREAMER. We train and test the selection outcomes using nested cross-validation and a support vector machine (SVM). We compared our results with the state-of-the-art subject-independent algorithms. Our results show an enhanced performance by accurately classifying human affection (i.e., based on valence and arousal) by 16%–27% compared to other studies. This work not only outperforms other subject-independent studies reported in the literature but also proposes an online analysis solution to affection recognition.

Social Cognition and Interaction Training (SCIT) versus Training in Affect Recognition (TAR) in patients with schizophrenia: A randomized controlled trial

IntroductionTraining in Affect Recognition (TAR) is a “targeted” and computer-aided program that has been shown to effectively attenuate facial affect recognition deficits and improve social functioning in patients with schizophrenia. Social Cognition and Interaction Training (SCIT) is a group “broad-based” intervention, that has also been shown to improve emotion recognition, theory of mind (ToM), and social functioning. To date, no study has compared the efficacy of two different social cognitive interventions. ObjectivesWe aim to compare the efficacy of TAR and SCIT on schizophrenia patients’ performance on facial affect recognition, theory of mind, attributional style and social functioning before, after treatment, and three months thereafter. MethodsOne hundred outpatients with a diagnosis of schizophrenia were randomly assigned to the TAR or SCIT condition and completed pre- (T0) and posttreatment (T1) assessments and a 3-month follow up (T2) of emotion recognition (ER-40), theory of mind (Hinting Task), attributional style (AIHQ) and social functioning (PSP). ResultsThe entire sample, receiving TAR or SCIT, showed improvements in theory of mind, attributional style, clinical symptoms and social functioning. This effect was maintained at three-months. The TAR intervention was more efficacious than the SCIT program in improving the recognition of facial emotions (ER-40). The TAR intervention also demonstrated a lower drop-out rate than the SCIT intervention. ConclusionsThere were improvements in social cognition, symptomatology and functioning of patients in the entire sample, receiving SCIT or TAR. Both TAR and SCIT appear as valuable treatments for people with schizophrenia and social cognitive deficits.

How Interpersonal Distance Between Avatar and Human Influences Facial Affect Recognition in Immersive Virtual Reality.

Purpose: The purpose of this study was to determine the optimal interpersonal distance (IPD) between humans and affective avatars in facial affect recognition in immersive virtual reality (IVR). The ideal IPD is the one in which the humans show the highest number of hits and the shortest reaction times in recognizing the emotions displayed by avatars. The results should help design future therapies to remedy facial affect recognition deficits.Methods: A group of 39 healthy volunteers participated in an experiment in which participants were shown 65 dynamic faces in IVR and had to identify six basic emotions plus neutral expression presented by the avatars. We decided to limit the experiment to five different distances: D1 (35 cm), D2 (55 cm), D3 (75 cm), D4 (95 cm), and D5 (115 cm), all belonging to the intimate and personal interpersonal spaces. Of the total of 65 faces, 13 faces were presented for each of the included distances. The views were shown at different angles: 50% in frontal view, 25% from the right profile, and 25% from the left profile. The order of appearance of the faces presented to each participant was randomized.Results: The overall success rate in facial emotion identification was 90.33%, being D3 the IPD with the best overall emotional recognition hits, although statistically significant differences could not be found between the IPDs. Consistent with results obtained in previous studies, identification rates for negative emotions were higher with increasing IPD, whereas the recognition task improved for positive emotions when IPD was closer. In addition, the study revealed irregular behavior in the facial detection of the emotion surprise.Conclusions: IVR allows us to reliably assess facial emotion recognition using dynamic avatars as all the IPDs tested showed to be effective. However, no statistically significant differences in facial emotion recognition were found among the different IPDs.

Characterization and interrelationships of theory of mind, socially competitive emotions and affective empathy in bipolar disorder.

Evidence shows impaired theory of mind (ToM) in patients with bipolar disorder (BD), yet research examining its cognitive and affective components simultaneously is sparse. Moreover, recognition of socially competitive 'fortune of others' emotions (e.g. envy/gloat) may be related to ToM, but has not been assessed in BD. Finally, if and how ToM and 'fortune of others' emotions relate to affective empathy in BD is currently unclear. This study aimed to address these points. 64 BD patients and 34 healthy controls completed the Yoni task, a visual task assessing first- and second-order cognitive and affective ToM as well as 'fortune of others' emotions. The Toronto Empathy Questionnaire was used to assess self-reported affective empathy. Patients with BD showed no deficits in cognitive and affective ToM or recognition of 'fortune of others' emotions. The ability to infer 'fortune of others' emotions correlated with several ToM measures, indicating that these functions are part of the same system. Patients with BD reported similar levels of affective empathy to healthy controls, and this was not related to ToM or 'fortune of others' emotions, suggesting that affective empathy represents a separate social domain. These findings highlight areas of spared social functioning in BD, which may be utilized in therapeutic strategies. Our results suggest theory of mind and empathy may represent areas of potentially spared cognitive functioning in BD. As many BD patients have experienced adversity during developmental periods in which theory of mind and empathy develop, our findings suggest that these abilities may be markers of resilience in the disorder. Our findings are important for the formulation of therapeutic interventions for BD, which may include considering practical ways that a patients' knowledge of intact ToM and empathy could be utilized to reduce self-stigma and promote self-efficacy, improved well-being and functioning.

Altered functional connectivity during face processing in children born with very low birth weight.

Structural brain alterations have been reported in key emotional face processing regions following preterm birth; however, few studies have investigated the functional networks underlying these processes in children born with very low birth weight (VLBW). Using magnetoencephalography (MEG), we examined the functional networks related to the implicit processing of happy and angry faces in 5-year-old VLBW (n = 28) and full-term (FT; n = 24) children. We found that VLBW children showed atypical recruitment of emotional face processing networks in theta (4–7 Hz) compared to FT children. VLBW children showed reduced theta connectivity during processing of angry faces only. This hypo-connected theta-band network was anchored in the left orbitofrontal and parietal regions, involved in the higher level processing of faces and emotion regulation. At the behavioural level, despite VLBW children performing within the normal range, FT children had significantly higher affect recognition scores. Our MEG results suggest a selective impairment in processing angry faces, which would negatively impact social functioning in VLBW children. In FT children, greater recruitment of this theta-band network was positively associated with improved affect recognition scores. Thus, our findings suggest an important role of theta oscillations in early face processing, deficits which may contribute to broader socio-emotional impairments in VLBW children.

Analysis of different affective state multimodal recognition approaches with missing data-oriented to virtual learning environments

In this work, the affective state of users in virtual learning environments is assessed/recognized in terms of continuous arousal and valence dimensions, making use of multimodal information (audio, text and video), whenever any of these modalities are available. In general, virtual learning environments where these three modalities are all the time, are not common; at some moments only the video modality is available, while in others only text or/and video and/or audio. Different approaches using feature-level fusion and decision-level fusion are proposed for multimodal recognition with missing data. Recognizing according to available modalities is studied following the ideas of dropout from neural networks and of variable input length from recurrent neural networks. This proposal is innovative because it represents emotions in the continuous space, which is not common in virtual education; and makes use of the available modalities in a virtual environment in a given moment, which is very common in virtual learning environments because the people are not speaking or writing all the time.

Social cognition and major depressive disorder: impact on psychosocial function and therapeutic opportunities

AimsThis poster aims to examine the impact of social cognitive deficits on psychosocial functioning in depressed patients, as well as summarise the utility of various evidence-based therapeutic interventions employed to target these deficits. The stated hypotheses were twofold: (1) that social cognitive impairment in major depressive disorder will correlate with poorer psychosocial functioning; and (2) that these deficits will respond to existing anti-depressant therapies.BackgroundSocial cognition is an important adaptive trait that incorporates the identification, perception and interpretation of socially relevant information from the external world. It is frequently affected in major depressive disorder such that depressed patienMethodA review of the existing literature was performed in order to test the stated hypotheses. Pertinent sources were identified via the MEDLINE, EMBASE, PsycINFO, PubMed, Scopus and Google Scholar databases. A total of 107 studies met inclusion criteria for review.ResultImpaired social cognitive performance in depressed patients correlated with poorer psychosocial functioning across the key domains of general cognitive functioning and quality of life. Many current anti-depressant therapies were found to have a normalising effect on the social cognitive abilities of depressed subjects, both at a neural and functional level. Anti-depressant medications, in particular citalopram and reboxetine, appeared to correct facial affect recognition deficits, while a psychotherapeutic approach demonstrated improvements in theory of mind and negative interpretive bias. Data relating to other common treatments, such as electroconvulsive therapy, are limited.ConclusionThe impact and treatment of social cognitive deficits in major depressive disorder is an important emerging field. The social cognitive deficits evident in depressed patients are sometimes subtle, but afford a significant functional impact. Additionally, it appears these impairments are at least partially reversible using anti-depressants or psychotherapy.

Genetically optimized Fuzzy C-means data clustering of IoMT-based biomarkers for fast affective state recognition in intelligent edge analytics

IoMT sensors such as wearables, moodables, ingestible sensors and trackers have the potential to provide a proactive approach to healthcare. But grouping, traversing and selectively tapping the IoMT data traffic and its immediacy makes data management & decision analysis a pressing issue. Evidently, the selection process for real-world, time-constrained health problems involves looking at multivariate time-series data generated simultaneously from various wearables resulting in data overload and accuracy issues. Computational intelligence of edge analytics can extend predictive capability by quickly turning digital biomarker data into actions for remote monitoring and trigger alarm during emergency incidents without relying on backend servers. But the pervasive generation of data streams from IoMT levies significant issues in data visualization and exploratory data analysis. This paper presents a genetically optimized Fuzzy C-means data clustering technique for affective state recognition on the edge. Clustering segregates the biomarker data in chunks and generates a summarized data for each subject which is then genetically optimized to avoid stagnation in local optima. A multi-level convolution neural network is finally used to classify the affective states into the baseline, stress and amusement categories. The model is evaluated on the publicly available WESAD dataset and compares favorably to state-of-the-art with less time complexity. It demonstrates the use of data clustering technique for numerosity reduction of real-time data streams in intelligent edge analytics which facilitates fast analysis of affective state of the user.

Use of affect context in dyadic interactions for continuous emotion recognition

Emotional dependencies play a crucial role in understanding complexities of dyadic interactions. Recent studies have shown that affect recognition tasks can benefit by the incorporation of a particular interaction’s context, however, the investigation of affect context in dyadic settings using neural network frameworks remains a complex and open problem. In this paper, we formulate the concept of dyadic affect context (DAC) and propose convolutional neural network (CNN) based architectures to model and incorporate DAC to improve continuous emotion recognition (CER) in dyadic scenarios. We begin by defining a CNN architecture for single-subject CER-based on speech and body motion data. We then introduce dyadic CER as a two-stage regression framework. Specifically, we propose two dyadic CNN architectures where cross-speaker affect contribution to the CER task is achieved by: (i) the fusion of cross-subject affect (FoA) or (ii) the fusion of cross-subject feature maps (FoM). Based on the preceding dyadic models, we finally propose a new Convolutional LSTM (ConvLSTM) model for the dyadic CER. ConvLSTM architecture captures local spectro-temporal correlations in speech and body motion as well as the long-term affect inter-dependencies between subjects. Our multimodal analysis demonstrates that modeling and incorporation of the DAC in the proposed CER models provide significant performance improvements on the USC CreativeIT database and the achieved results compare favorably to the state-of-the-art.

Product innovation concept generation based on deep learning and Kansei engineering

Industrial designers often present their initial concepts as design sketches. Rapid creation of new product conceptual images that meet users’ affective preferences remains challenging in real design environments. However, few published works in affective design directly assist industrial designers in creating product conceptual images. Thus, we propose a product concept generation approach framework based on deep learning and Kansei engineering (PCGA-DLKE) to assist industrial designers. Our work focuses on dataset collection, pre-processing, affective preferences recognition, conceptual image generation model and product style transfer networks. To mark users’ affective preferences, we established an affective recognition model by Kansei engineering and deep convolutional neural networks. To address the product conceptual image generation problem, we proposed a product design GAN model (PD-GAN), generating product conceptual images with affective preferences. An improved fast neural style transfer network was successfully trained to meet users’ style preferences. This study aims to assist industrial designers in finding innovative concepts with affective preference. The Kansei evaluation shows that the innovation of the new product concept has been enhanced, indicating that the approach can better assist industrial designers in creating designs that meet users’ emotional needs. Hand drill design and bicycle helmet design are taken as a case study.