Sometimes memory misleads: Variants of the error-speed effect strengthen the evidence for systematically misleading memory signals in recognition memory.

The error-speed effect describes the observation that the speed of recognition errors in a first binary recognition task predicts the response accuracy in a subsequent two-alternative forced-choice (2AFC) task that comprises the erroneously judged items of the first task. So far, the effect has been primarily explained by the assumption that some error responses result from misleading memory evidence. However, it is also possible that the effect arises because participants remember and use their response times from the binary task to solve the 2AFC task. Furthermore, the phenomenon is quite new and its robustness or generalizability across other recognition tasks (e.g., a confidence-rating task) remains to be demonstrated. The aim of the present study is to address these limitations by introducing a new variant of the error-speed effect, replacing the 2AFC task with a confidence-rating task (Experiment 1), and by reversing task order (Experiment 2) to test whether participants employ a response-time strategy. In both experiments, we collected data using a sequential probability ratio t-test procedure and found evidence in favor of the hypothesis that the speed of binary recognition errors predicts confidence ratings for the same stimulus. These results attest to the robustness and generalizability of the error-speed effect and reveal that at least some errors must be due to systematically misleading memory evidence.

EXPRESS:The "Good is Up" Metaphoric Effects on Recognition: True for Source Guessing but False for Item Memory.

The "good is up" metaphor which links valence and verticality was found to influence affective judgement and to direct attention (Meier & Robinson, 2004), but its effects on memory remain unclear with contradictory research findings. In order to provide a more accurate assessment of memory components involved in recognition, such as item memory and source guessing biases, a standard source monitoring paradigm (Johnson et al., 1993) was applied in this research. A series of three experiments, provided a conceptual replication and extension of Crawford et al's (2014) Experiment 2 and yielded a consistent result pattern suggesting that the "good is up" metaphor biases participants' guessing of source location. That is, when source memory failed, participants were more inclined to guess the "up" location for positive items versus "down" location for negative items. It did, however, not affect source memory or item memory for valenced stimuli learned from metaphor-congruent versus incongruent locations (i.e., no metaphor-(in)congruent effects in memory). We suggest that the "good is up" metaphor may affect cognitive processes in a more subtle way than originally suggested.

A Feature-Space Theory of the Production Effect in Recognition.

Mathematical models explaining production effects assume that production leads to the encoding of additional features, such as phonological ones. This improves memory with a combination of encoding strength and feature distinctiveness, implementing aspects of propositional theories. However, it is not clear why production differs from other manipulations such as study time and spaced repetition, which are also thought to influence strength. Here we extend attentional subsetting theory and propose an explanation based on the dimensionality of feature spaces. Specifically, we suggest phonological features are drawn from a compact feature space. Deeper features are sparsely subselected from a larger subspace. Algebraic and numerical solutions shed light on several findings, including the dependency of production effects on how other list items are encoded (differing from other strength factors) and the production advantage even for homophones. This places production within a continuum of strength-like manipulations that differ in terms of the feature subspaces they operate upon and leads to novel predictions based on direct manipulations of feature-space properties.

The disconnect between metamemory and memory for emotional images.

Emotional information is reliably predicted to be remembered better than neutral information, and this has been found for words, images, and facial expressions. However, many studies find that these judgments of learning (JOLs) are not predictive of memory performance (e.g. [Hourihan, K. L. (2020). Misleading emotions: Judgments of learning overestimate recognition of negative and positive emotional images. Cognition and Emotion, 34(4), 771-782. https://doi.org/10.1080/02699931.2019.1682972]). The present study investigates and rules out numerous potential causes for this discrepancy between memory predictions and performance, including (1) reactivity to making JOLs, (2) idiosyncrasies of specific images used, (3), type of memory test, and (4) effects of fluency. Three additional experiments investigate whether JOLs can become more predictive of memory performance, either by experience with the task or by manipulating prior beliefs about memory for emotional images. In all experiments, we found the same effect: Emotional images are inaccurately predicted to be remembered better than neutral images. The results suggest that emotion is used as a heuristic for learning, resulting in low metamnemonic accuracy for emotional stimuli.

Early-life manipulation of the serotonergic system exacerbates the harmful effects of sleep deprivation on cognitive functions.

Serotonin is a monoamine neurotransmitter that plays a main role in regulating physiological and cognitive functions. Serotonergic system dysfunction is involved in the etiology of various psychiatric and neurological disorders. Therefore, the present study was designed to investigate the effects of early-life serotonin depletion on cognitive disorders caused by sleep deprivation. Serotonin was depleted by para-chlorophenylalanine (PCPA, 100 mg/kg, s.c.) at postnatal days 10-20, followed by sleep deprivation-induced through the multiple platform apparatus for 24 h at PND 60. After the examination of the novel object recognition and passive avoidance memories, the hippocampi and prefrontal cortex were dissected to examine the brain-derived neurotrophic factor (BDNF) mRNA expression by PCR. Our findings showed that postnatal serotonin depletion and sleep deprivation impaired the novel object recognition and passive avoidance memories and changed the BDNF levels. In the same way, the serotonin depletion in early life before sleep deprivation exacerbated the harmful effects of sleep deprivation on cognitive function and BDNF levels. It can be claimed that the serotonergic system plays a main role in the modulation of sleep and cognitive functions.

Dissecting the contribution of human chromosome 21 syntenic regions to recognition memory processes in adult and aged mouse models of Down syndrome.

Background: Trisomy of human chromosome 21 (Hsa21) results in a constellation of features known as Down syndrome (DS), the most common genetic form of intellectual disability. Hsa21 is orthologous to three regions in the mouse genome on mouse chromosome 16 (Mmu16), Mmu17 and Mmu10. We investigated genotype-phenotype relationships by assessing the contribution of these three regions to memory function and age-dependent cognitive decline, using three mouse models of DS, Dp1Tyb, Dp(17)3Yey, Dp(10)2Yey, that carry an extra copy of the Hsa21-orthologues on Mmu16, Mmu17 and Mmu10, respectively. Hypothesis: Prior research on cognitive function in DS mouse models has largely focused on models with an extra copy of the Mmu16 region and relatively little is known about the effects of increased copy number on Mmu17 and Mmu10 on cognition and how this interacts with the effects of aging. As aging is is a critical contributor to cognitive and psychiatric changes in DS, we hypothesised that ageing would differentially impact memory function in Dp1Tyb, Dp(17)3Yey, and Dp(10)2Yey, models of DS. Methods: Young (12-13 months and old (18-20 months mice Dp1Tyb, Dp(17)3Yey and Dp(10)2Yey mice were tested on a battery of object recognition memory test that assessed object novelty detection, novel location detection and associative object-in place memory. Following behavioral testing, hippocampal and frontal cortical tissue was analysed for expression of glutamatergic receptor proteins using standard immunoblot techniques. Results: Young (12-13 months and old (18-20 months mice Dp1Tyb, Dp(17)3Yey and Dp(10)2Yey mice were tested on a battery of object recognition memory test that assessed object novelty detection, novel location detection and associative object-in place memory. Following behavioral testing, hippocampal and frontal cortical tissue was analysed for expression of glutamatergic receptor proteins using standard immunoblot techniques. Conclusion: Our results show that distinct Hsa21-orthologous regions contribute differentially to cognitive dysfunction in DS mouse models and that aging interacts with triplication of Hsa21-orthologous genes on Mmu10.

Differential Alterations in the Expression of AMPA Receptor and Its Trafficking Proteins in the Hippocampus Are Associated with Recognition Memory Impairment in the Rotenone-Parkinson's Disease Mouse Model: Neuroprotective Role of Bacopa monnieri Extract CDRI 08.

Parkinson's disease (PD), an age-associated neurodegenerative motor disorder, is associated with dementia and cognitive decline. However, the precise molecular insight into PD-induced cognitive decline is not fully understood. Here, we have investigated the possible alterations in the expression of glutamate receptor and its trafficking/scaffolding/regulatory proteins underlying the memory formation and neuroprotective effects of a specialized Bacopa monnieri extract, CDRI-08 (BME) in the hippocampus of the rotenone-induced PD mouse model. Our Western blotting and qRT-PCR data reveal that the PD-induced recognition memory decline is associated with significant upregulation of the AMPA receptor subunit GluR1 and downregulation of GluR2 subunit genes in the hippocampus of rotenone-affected mice as compared to the vehicle control. Further, expressions of the trafficking proteins are significantly upregulated in the hippocampus of rotenone-affected mice compared to the vehicle control. Our results also reveal that the above alterations in the hippocampus are associated with similar expression patterns of total CREB, pCREB, and BDNF. BME (CDRI-08, 200 mg/kg BW) reverses the expression of AMPA receptor subunits, their trafficking proteins differentially, and the transcriptional modulatory proteins depending on whether the BME treatment was given before or after the rotenone treatment. Our data suggest that expression of the above genes is significantly reversed in the BME pre-treated mice subjected to rotenone treatment towards their levels in the control mice compared to its treatment after rotenone administration. Our results provide the possible molecular basis underlying the rotenone-induced recognition memory decline, conditions mimicking the PD symptoms in mouse model and neuroprotective action of bacoside A and bacoside B (58%)-enriched Bacopa monnieri extract (BME) in the hippocampus.

Early spatial recognition memory deficits in 5XFAD female mice are associated with disruption of prefrontal parvalbumin neurons.

Women have a two-fold increased risk of developing Alzheimer's disease (AD) than men, yet the underlying mechanisms of this sex-specific vulnerability remain unknown. Here, we aimed at determining in the 5XFAD mouse model whether deficits in prefrontal-dependent cognitive functions, which are impacted in the preclinical stages of AD, appear earlier in females, and whether these cognitive deficits are associated with alterations in the activity of prefrontal parvalbumin (PV)-neurons that regulate prefrontal circuits activity. We observed that 3.5-month-old 5XFAD females, but not males, display impairments in spatial short-term recognition memory, a function that relies on the integrity of the prefrontal cortex. Hippocampal-dependent cognitive functions were intact in both sexes. We then observed that 5XFAD females have more prefrontal PV neurons expressing the marker of chronic activity FosB; this was inversely correlated with prefrontal-dependent cognitive performances. Our findings show for the first time sex-specific, early deregulation of prefrontal PV neurons activity, which is associated with early appearance of prefrontal-dependent cognitive functions in 5XFAD females providing a potential novel mechanism to the increased risk to AD in females.

Treadmill exercise prevents recognition memory impairment in VD rat model and enhancement of hippocampal structural synaptic plasticity.

OBJECTIVE: In vascular dementia (VD), memory impairment caused by the damage of synaptic plasticity is the most prominent feature that afflicts patients and their families. Treadmill exercise has proven beneficial for memory by enhancing synaptic plasticity in animal models including stroke, dementia, and mental disorders. The aim of this study was to examine the effects of treadmill exercise on recognition memory and structural synaptic plasticity in VD rat model. METHODS: Male Sprague-Dawley rats were randomly assigned into four groups: control group (C group, n = 6), vascular dementia group (VD group, n = 6), treadmill exercise and vascular dementia group (Exe-VD group, n = 6), and treadmill exercise group (Exe group, n = 6). Four-week treadmill exercise was performed in the Exe-VD and Exe groups. Then, the common carotid arteries of rats in the VD and Exe-VD groups were identified to establish the VD model. Behavior tests (open-field test and novel recognition memory test) were adopted to evaluate anxiety-like behavior and recognition memory. Transmission electron microscopy and Golgi staining were performed to observe synaptic ultrastructure and spine density in the hippocampus. RESULTS: Our study demonstrated that VD rat exhibited significantly anxiety-like behavior and recognition impairment (p < .01), while treadmill exercise significantly alleviated anxiety-like behavior and improved recognition memory in VD rat (p < .01). Transmission electron microscopy revealed that hippocampal synapse numbers were significantly decreased in the VD group compared to the control group (p < .05). These alterations were reversed by treadmill exercise, and the rats exhibited healthier synaptic ultrastructure, including significantly increased synapse (p < .05). Meanwhile, golgi staining revealed that the spine numbers of the hippocampus were significantly decreased in the VD group compared to the control group (p < .05). When compared with the VD group, hippocampal spine numbers were significantly increased in the Exe-VD group (p < .05). CONCLUSION: The improvement of VD-associated recognition memory by treadmill exercises is associated with enhanced structural synaptic plasticity in VD rat model.

Delineating the molecular mechanisms of hippocampal neurotoxicity induced by chronic administration of synthetic cannabinoid AB-FUBINACA in mice.

Chronic use of synthetic cannabinoids (SCs) has been associated with cognitive and behavioural deficits and an increased risk of neuropsychiatric disorders. The underlying molecular and cellular mechanisms of the neurotoxic effects of long-term use of SCs have not been well investigated in the literature. Herein, we evaluated the in vivo effects of chronic administration of AB-FUBINACA on the hippocampus in mice. Our results revealed that the administration of AB-FUBINACA induced a significant impairment in recognition memory associated with histopathological changes in the hippocampus. These findings were found to be correlated with increased level of oxidative stress, neuroinflammation, and apoptosis markers, and reduced expression of brain-derived neurotrophic factor (BDNF), which plays an essential role in modulating synaptic plasticity integral for promoting learning and memory in the hippocampus. Additionally, we showed that AB-FUBINACA significantly decreased the expression of NR1, an important functional subunit of glutamate/NMDA receptors and closely implicated in the development of toxic psychosis. These findings shed light on the long-term neurotoxic effects of SCs on hippocampus and the underlying mechanisms of these effects. This study provided new targets for possible medical interventions to improve the treatment guidelines for SCs addiction.

The classification accuracy of Warrington's recognition memory test (words) as a performance validity Test in a neurorehabilitation setting.

This study was designed to evaluate the classification accuracy of the Warrington's Recognition Memory Test (RMT) in 167 patients (97 or 58.1% men; MAge = 40.4; MEducation= 13.8) medically referred for neuropsychological evaluation against five psychometrically defined criterion groups. At the optimal cutoff (≤42), the RMT produced an acceptable combination of sensitivity (.36-.60) and specificity (.85-.95), correctly classifying 68.4-83.3% of the sample. Making the cutoff more conservative (≤41) improved specificity (.88-.95) at the expense of sensitivity (.30-.60). Lowering the cutoff to ≤40 achieved uniformly high specificity (.91-.95) but diminished sensitivity (.27-.48). RMT scores were unrelated to lateral dominance, education, or gender. The RMT was sensitive to a three-way classification of performance validity (Pass/Borderline/Fail), further demonstrating its discriminant power. Despite a notable decline in research studies focused on its classification accuracy within the last decade, the RMT remains an effective free-standing PVT that is robust to demographic variables. Relatively low sensitivity is its main liability. Further research is needed on its cross-cultural validity (sensitivity to limited English proficiency).

Object Recognition Memory Deficits in ADHD: A Meta-analysis.

Object recognition memory allows us to identify previously seen objects. This type of declarative memory is a primary process for learning. Despite its crucial role in everyday life, object recognition has received far less attention in ADHD research compared to verbal recognition memory. In addition to the existence of a small number of published studies, the results have been inconsistent, possibly due to the diversity of tasks used to assess recognition memory. In the present meta-analysis, we have collected studies from Web of Science, Scopus, PubMed, and Google Scholar databases up to May 2023. We have compiled studies that assessed visual object recognition memory with specific visual recognition tests (sample-match delayed tasks) in children and adolescents diagnosed with ADHD. A total of 28 studies with 1619 participants diagnosed with ADHD were included. The studies were assessed for risk of bias using the Quadas-2 tool and for each study, Cohen's d was calculated to estimate the magnitude of the difference in performance between groups. As a main result, we have found a worse recognition memory performance in ADHD participants when compared to their matched controls (overall Cohen's d ~ 0.492). We also observed greater heterogeneity in the magnitude of this deficit among medicated participants compared to non-medicated individuals, as well as a smaller deficit in studies with a higher proportion of female participants. The magnitude of the object recognition memory impairment in ADHD also seems to depend on the assessment method used.

Single-dose methamphetamine administration impairs ORM retrieval in mice via excessive DA-mediated inhibition of PrLGlu activity.

Methamphetamine (METH), an abused psychostimulant, impairs cognition through prolonged or even single-dose exposure, but animal experiments have shown contradictory effects on memory deficits. In this study we investigated the effects and underlying mechanisms of single-dose METH administration on the retrieval of object recognition memory (ORM) in mice. We showed that single-dose METH administration (2 mg/kg, i.p.) significantly impaired ORM retrieval in mice. Fiber photometry recording in METH-treated mice revealed that the activity of prelimbic cortex glutamatergic neurons (PrLGlu) was significantly reduced during ORM retrieval. Chemogenetic activation of PrLGlu or glutamatergic projections from ventral CA1 to PrL (vCA1Glu-PrL) rescued ORM retrieval impairment. Fiber photometry recording revealed that dopamine (DA) levels in PrL of METH-treated mice were significantly increased, and micro-infusion of the D2 receptor (D2R) antagonist sulpiride (0.25 µg/side) into PrL rescued ORM retrieval impairment. Whole-cell recordings in brain slices containing the PrL revealed that PrLGlu intrinsic excitability and basal glutamatergic synaptic transmission were significantly reduced in METH-treated mice, and the decrease in intrinsic excitability was reversed by micro-infusion of Sulpiride into PrL in METH-treated mice. Thus, the impaired ORM retrieval caused by single-dose METH administration may be attributed to reduced PrLGlu activity, possibly due to excessive DA activity on D2R. Selective activation of PrLGlu or vCA1Glu-PrL may serve as a potential therapeutic strategy for METH-induced cognitive dysfunction.

Face recognition's practical relevance: Social bonds, not social butterflies.

Research on individual differences in face recognition has provided important foundational insights: their broad range, cognitive specificity, strong heritability, and resilience to change. Elusive, however, has been the key issue of practical relevance: do these individual differences correlate with aspects of life that go beyond the recognition of faces, per se? Though often assumed, especially in social realms, such correlates remain largely theoretical, without empirical support. Here, we investigate an array of potential social correlates of face recognition. We establish social relationship quality as a reproducible correlate. This link generalises across face recognition tasks and across independent samples. In contrast, we detect no robust association with the sheer quantity of social connections, whether measured directly via number of social contacts or indirectly via extraversion-related personality indices. These findings document the existence of a key social correlate of face recognition and provide some of the first evidence to support its practical relevance. At the same time, they challenge the naive assumption that face recognition relates equally to all social outcomes. In contrast, they suggest a focused link of face recognition to the quality, not quantity, of one's social connections.

Multifactorial effects of short duration early exposure low intensity magnetic field stimulation in Streptozotocin induced Alzheimer's disease rat model.

Alzheimer's disease (AD) is an approaching, progressive public health crisis which presently lacks an effective treatment. Various non-invasive novel therapies like repetitive transcranial magnetic stimulation have shown potential to improve cognitive performance in AD patients. In the present study, the effect of extremely low intensity magnetic field (MF) stimulation on neurogenesis and cortical electrical activity was explored. Adult Wistar rats were divided into Sham, AD and AD + MF groups. Streptozotocin (STZ) was injected intracerebroventricularly, at a dose of 3 mg/kg body weight for developing AD model. The AD rats were then exposed to MF (17.96 µT) from 8th day of STZ treatment until 15th day, followed by cognitive assessments and electrocortical recording. In brain tissue samples, cresyl violet staining and BrdU immunohistochemistry were done. MF exposure, improved passive avoidance and recognition memory, attenuated neuronal degeneration and enhanced cell proliferation (BrdU positive cells) in comparison to AD rats. It also significantly restores delta wave power from frontal lobe. Our results suggest that early-stage MF exposure could be an asset for AD research and open new avenues in slowing down the progression of Alzheimer's disease.

The late positive event-related potential component is time locked to the decision in recognition memory tasks.

Two event-related potential (ERP) components are commonly observed in recognition memory tasks: the Frontal Negativity (FN400) and the Late Positive Component (LPC). These components are widely interpreted as neural correlates of familiarity and recollection, respectively. However, the interpretation of LPC effects is complicated by inconsistent results regarding the timing of ERP amplitude differences. There are also mixed findings regarding how LPC amplitudes covary with decision confidence. Critically, LPC effects have almost always been measured using fixed time windows relative to memory probe stimulus onset, yet it has not been determined whether LPC effects are time locked to the stimulus or the recognition memory decision. To investigate this, we analysed a large (n = 132) existing dataset recorded during recognition memory tasks with old/new decisions followed by post-decisional confidence ratings. We used ERP deconvolution to disentangle contributions to LPC effects (defined as differences between hits and correct rejections) that were time locked to either the stimulus or the vocal old/new response. We identified a left-lateralised parietal LPC effect that was time locked to the vocal response rather than probe stimulus onset. We also isolated a response-locked, midline parietal ERP correlate of confidence that influenced measures of LPC amplitudes at left parietal electrodes. Our findings demonstrate that, contrary to widespread assumptions, the LPC effect is time locked to the recognition memory decision and is best measured using response-locked ERPs. By extension, differences in response time distributions across conditions of interest may lead to substantial measurement biases when analysing stimulus-locked ERPs. Our findings highlight important confounding factors that further complicate the interpretation of existing stimulus-locked LPC effects as neural correlates of recollection. We recommend that future studies adopt our analytic approach to better isolate LPC effects and their sensitivity to manipulations in recognition memory tasks.

The hippocampus supports the representation of abstract concepts: Implications for the study of recognition memory.

Words, unlike images, are symbolic representations. The associative details inherent within a word's meaning and the visual imagery it generates, are inextricably connected to the way words are processed and represented. It is well recognised that the hippocampus associatively binds components of a memory to form a lasting representation, and here we show that the hippocampus is especially sensitive to abstract word processing. Using fMRI during recognition, we found that the increased abstractness of words produced increased hippocampal activation regardless of memory outcome. Interestingly, word recollection produced hippocampal activation regardless of word content, while the parahippocampal cortex was sensitive to concreteness of word representations, regardless of memory outcome. We reason that the hippocampus has assumed a critical role in the representation of uncontextualized abstract word meaning, as its information-binding ability allows the retrieval of the semantic and visual associates that, when bound together, generate the abstract concept represented by word symbols. These insights have implications for research on word representation, memory, and hippocampal function, perhaps shedding light on how the human brain has adapted to encode and represent abstract concepts.

Seeing what you believe: recognition memory for evolutionary tree structure is affected by students' misconceptions.

Peoples' recognition memory for pictorial stimuli is extremely good. Even complex scientific visualisations are recognised with a high degree of accuracy. The present research examined recognition memory for the branching structure of evolutionary trees. This is an educationally consequential topic due to the potential for contamination from students' misconceptions. The authors created six pairs of scientifically accurate and structurally identical evolutionary trees that differed in whether they included a taxon that cued a misconception in memory. As predicted, Experiment 1 found that (a) college students (N = 90) had better memory for each of the six tree structures when a neutral taxon (M = 0.73) rather than a misconception-cuing taxon (M = 0.64) was included in the tree, and (b) recognition memory was significantly above chance for both sets of trees. Experiment 2 ruled out an alternative hypothesis based on the possibility that 8-12 sec was not enough time for students to encode the relationships depicted in the trees. The authors consider implications of these results for using evolutionary trees to better communicate scientific information. This is important because these trees provide information that is relevant for everyday life.

Phase locking of hippocampal CA3 neurons to distal CA1 theta oscillations selectively predicts memory performance.

How the coordination of neuronal spiking and brain rhythms between hippocampal subregions supports memory function remains elusive. We studied the interregional coordination of CA3 neuronal spiking with CA1 theta oscillations by recording electrophysiological signals along the proximodistal axis of the hippocampus in rats that were performing a high-memory-demand recognition memory task adapted from humans. We found that CA3 population spiking occurs preferentially at the peak of distal CA1 theta oscillations when memory was tested but only when previously encountered stimuli were presented. In addition, decoding analyses revealed that only population cell firing of proximal CA3 together with that of distal CA1 can predict performance at test in the present non-spatial task. Overall, our work demonstrates an important role for the synchronization of CA3 neuronal activity with CA1 theta oscillations during memory testing.

Detecting recollection: Human evaluators can successfully assess the veracity of others' memories.

Humans have the highly adaptive ability to learn from others' memories. However, because memories are prone to errors, in order for others' memories to be a valuable source of information, we need to assess their veracity. Previous studies have shown that linguistic information conveyed in self-reported justifications can be used to train a machine-learner to distinguish true from false memories. But can humans also perform this task, and if so, do they do so in the same way the machine-learner does? Participants were presented with justifications corresponding to Hits and False Alarms and were asked to directly assess whether the witness's recognition was correct or incorrect. In addition, participants assessed justifications' recollective qualities: their vividness, specificity, and the degree of confidence they conveyed. Results show that human evaluators can discriminate Hits from False Alarms above chance levels, based on the justifications provided per item. Their performance was on par with the machine learner. Furthermore, through assessment of the perceived recollective qualities of justifications, participants were able to glean more information from the justifications than they used in their own direct decisions and than the machine learner did.